new bert
Browse filesThis view is limited to 50 files because it contains too many changes.
See raw diff
- bert-master/bert-master.gitignore +116 -0
- bert-master/bert-master/CONTRIBUTING.md +31 -0
- bert-master/bert-master/LICENSE +202 -0
- bert-master/bert-master/README.md +1117 -0
- bert-master/bert-master/__init__.py +15 -0
- bert-master/bert-master/create_pretraining_data.py +469 -0
- bert-master/bert-master/extract_features.py +419 -0
- bert-master/bert-master/modeling.py +986 -0
- bert-master/bert-master/modeling_test.py +277 -0
- bert-master/bert-master/multilingual.md +303 -0
- bert-master/bert-master/optimization.py +174 -0
- bert-master/bert-master/optimization_test.py +48 -0
- bert-master/bert-master/predicting_movie_reviews_with_bert_on_tf_hub.ipynb +1231 -0
- bert-master/bert-master/requirements.txt +2 -0
- bert-master/bert-master/run_classifier.py +981 -0
- bert-master/bert-master/run_classifier_with_tfhub.py +314 -0
- bert-master/bert-master/run_pretraining.py +493 -0
- bert-master/bert-master/run_squad.py +1283 -0
- bert-master/bert-master/sample_text.txt +33 -0
- bert-master/bert-master/tokenization.py +399 -0
- bert-master/bert-master/tokenization_test.py +137 -0
- dark-bert-master/dark-bert-master/LICENSE +201 -0
- dark-bert-master/dark-bert-master/README.md +14 -0
- dark-bert-master/dark-bert-master/darkbert.py +151 -0
- dark-bert-master/dark-bert-master/requirements.txt +0 -0
- dark-secrets-of-BERT-master/dark-secrets-of-BERT-master.gitignore +125 -0
- dark-secrets-of-BERT-master/dark-secrets-of-BERT-master/.circleci/config.yml +29 -0
- dark-secrets-of-BERT-master/dark-secrets-of-BERT-master/.github/stale.yml +17 -0
- dark-secrets-of-BERT-master/dark-secrets-of-BERT-master/LICENSE +202 -0
- dark-secrets-of-BERT-master/dark-secrets-of-BERT-master/MANIFEST.in +1 -0
- dark-secrets-of-BERT-master/dark-secrets-of-BERT-master/README.md +30 -0
- dark-secrets-of-BERT-master/dark-secrets-of-BERT-master/docker/Dockerfile +7 -0
- dark-secrets-of-BERT-master/dark-secrets-of-BERT-master/docs/imgs/warmup_constant_schedule.png +0 -0
- dark-secrets-of-BERT-master/dark-secrets-of-BERT-master/docs/imgs/warmup_cosine_hard_restarts_schedule.png +0 -0
- dark-secrets-of-BERT-master/dark-secrets-of-BERT-master/docs/imgs/warmup_cosine_schedule.png +0 -0
- dark-secrets-of-BERT-master/dark-secrets-of-BERT-master/docs/imgs/warmup_cosine_warm_restarts_schedule.png +0 -0
- dark-secrets-of-BERT-master/dark-secrets-of-BERT-master/docs/imgs/warmup_linear_schedule.png +0 -0
- dark-secrets-of-BERT-master/dark-secrets-of-BERT-master/examples/extract_features.py +297 -0
- dark-secrets-of-BERT-master/dark-secrets-of-BERT-master/examples/lm_finetuning/README.md +64 -0
- dark-secrets-of-BERT-master/dark-secrets-of-BERT-master/examples/lm_finetuning/finetune_on_pregenerated.py +333 -0
- dark-secrets-of-BERT-master/dark-secrets-of-BERT-master/examples/lm_finetuning/pregenerate_training_data.py +302 -0
- dark-secrets-of-BERT-master/dark-secrets-of-BERT-master/examples/lm_finetuning/simple_lm_finetuning.py +642 -0
- dark-secrets-of-BERT-master/dark-secrets-of-BERT-master/examples/run_classifier.py +1047 -0
- dark-secrets-of-BERT-master/dark-secrets-of-BERT-master/examples/run_gpt2.py +133 -0
- dark-secrets-of-BERT-master/dark-secrets-of-BERT-master/examples/run_openai_gpt.py +274 -0
- dark-secrets-of-BERT-master/dark-secrets-of-BERT-master/examples/run_squad.py +1098 -0
- dark-secrets-of-BERT-master/dark-secrets-of-BERT-master/examples/run_swag.py +551 -0
- dark-secrets-of-BERT-master/dark-secrets-of-BERT-master/examples/run_transfo_xl.py +153 -0
- dark-secrets-of-BERT-master/dark-secrets-of-BERT-master/hubconf.py +187 -0
- dark-secrets-of-BERT-master/dark-secrets-of-BERT-master/notebooks/Comparing-TF-and-PT-models-MLM-NSP.ipynb +0 -0
bert-master/bert-master.gitignore
ADDED
|
@@ -0,0 +1,116 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# Initially taken from Github's Python gitignore file
|
| 2 |
+
|
| 3 |
+
# Byte-compiled / optimized / DLL files
|
| 4 |
+
__pycache__/
|
| 5 |
+
*.py[cod]
|
| 6 |
+
*$py.class
|
| 7 |
+
|
| 8 |
+
# C extensions
|
| 9 |
+
*.so
|
| 10 |
+
|
| 11 |
+
# Distribution / packaging
|
| 12 |
+
.Python
|
| 13 |
+
build/
|
| 14 |
+
develop-eggs/
|
| 15 |
+
dist/
|
| 16 |
+
downloads/
|
| 17 |
+
eggs/
|
| 18 |
+
.eggs/
|
| 19 |
+
lib/
|
| 20 |
+
lib64/
|
| 21 |
+
parts/
|
| 22 |
+
sdist/
|
| 23 |
+
var/
|
| 24 |
+
wheels/
|
| 25 |
+
*.egg-info/
|
| 26 |
+
.installed.cfg
|
| 27 |
+
*.egg
|
| 28 |
+
MANIFEST
|
| 29 |
+
|
| 30 |
+
# PyInstaller
|
| 31 |
+
# Usually these files are written by a python script from a template
|
| 32 |
+
# before PyInstaller builds the exe, so as to inject date/other infos into it.
|
| 33 |
+
*.manifest
|
| 34 |
+
*.spec
|
| 35 |
+
|
| 36 |
+
# Installer logs
|
| 37 |
+
pip-log.txt
|
| 38 |
+
pip-delete-this-directory.txt
|
| 39 |
+
|
| 40 |
+
# Unit test / coverage reports
|
| 41 |
+
htmlcov/
|
| 42 |
+
.tox/
|
| 43 |
+
.nox/
|
| 44 |
+
.coverage
|
| 45 |
+
.coverage.*
|
| 46 |
+
.cache
|
| 47 |
+
nosetests.xml
|
| 48 |
+
coverage.xml
|
| 49 |
+
*.cover
|
| 50 |
+
.hypothesis/
|
| 51 |
+
.pytest_cache/
|
| 52 |
+
|
| 53 |
+
# Translations
|
| 54 |
+
*.mo
|
| 55 |
+
*.pot
|
| 56 |
+
|
| 57 |
+
# Django stuff:
|
| 58 |
+
*.log
|
| 59 |
+
local_settings.py
|
| 60 |
+
db.sqlite3
|
| 61 |
+
|
| 62 |
+
# Flask stuff:
|
| 63 |
+
instance/
|
| 64 |
+
.webassets-cache
|
| 65 |
+
|
| 66 |
+
# Scrapy stuff:
|
| 67 |
+
.scrapy
|
| 68 |
+
|
| 69 |
+
# Sphinx documentation
|
| 70 |
+
docs/_build/
|
| 71 |
+
|
| 72 |
+
# PyBuilder
|
| 73 |
+
target/
|
| 74 |
+
|
| 75 |
+
# Jupyter Notebook
|
| 76 |
+
.ipynb_checkpoints
|
| 77 |
+
|
| 78 |
+
# IPython
|
| 79 |
+
profile_default/
|
| 80 |
+
ipython_config.py
|
| 81 |
+
|
| 82 |
+
# pyenv
|
| 83 |
+
.python-version
|
| 84 |
+
|
| 85 |
+
# celery beat schedule file
|
| 86 |
+
celerybeat-schedule
|
| 87 |
+
|
| 88 |
+
# SageMath parsed files
|
| 89 |
+
*.sage.py
|
| 90 |
+
|
| 91 |
+
# Environments
|
| 92 |
+
.env
|
| 93 |
+
.venv
|
| 94 |
+
env/
|
| 95 |
+
venv/
|
| 96 |
+
ENV/
|
| 97 |
+
env.bak/
|
| 98 |
+
venv.bak/
|
| 99 |
+
|
| 100 |
+
# Spyder project settings
|
| 101 |
+
.spyderproject
|
| 102 |
+
.spyproject
|
| 103 |
+
|
| 104 |
+
# Rope project settings
|
| 105 |
+
.ropeproject
|
| 106 |
+
|
| 107 |
+
# mkdocs documentation
|
| 108 |
+
/site
|
| 109 |
+
|
| 110 |
+
# mypy
|
| 111 |
+
.mypy_cache/
|
| 112 |
+
.dmypy.json
|
| 113 |
+
dmypy.json
|
| 114 |
+
|
| 115 |
+
# Pyre type checker
|
| 116 |
+
.pyre/
|
bert-master/bert-master/CONTRIBUTING.md
ADDED
|
@@ -0,0 +1,31 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# How to Contribute
|
| 2 |
+
|
| 3 |
+
BERT needs to maintain permanent compatibility with the pre-trained model files,
|
| 4 |
+
so we do not plan to make any major changes to this library (other than what was
|
| 5 |
+
promised in the README). However, we can accept small patches related to
|
| 6 |
+
re-factoring and documentation. To submit contributes, there are just a few
|
| 7 |
+
small guidelines you need to follow.
|
| 8 |
+
|
| 9 |
+
## Contributor License Agreement
|
| 10 |
+
|
| 11 |
+
Contributions to this project must be accompanied by a Contributor License
|
| 12 |
+
Agreement. You (or your employer) retain the copyright to your contribution;
|
| 13 |
+
this simply gives us permission to use and redistribute your contributions as
|
| 14 |
+
part of the project. Head over to <https://cla.developers.google.com/> to see
|
| 15 |
+
your current agreements on file or to sign a new one.
|
| 16 |
+
|
| 17 |
+
You generally only need to submit a CLA once, so if you've already submitted one
|
| 18 |
+
(even if it was for a different project), you probably don't need to do it
|
| 19 |
+
again.
|
| 20 |
+
|
| 21 |
+
## Code reviews
|
| 22 |
+
|
| 23 |
+
All submissions, including submissions by project members, require review. We
|
| 24 |
+
use GitHub pull requests for this purpose. Consult
|
| 25 |
+
[GitHub Help](https://help.github.com/articles/about-pull-requests/) for more
|
| 26 |
+
information on using pull requests.
|
| 27 |
+
|
| 28 |
+
## Community Guidelines
|
| 29 |
+
|
| 30 |
+
This project follows
|
| 31 |
+
[Google's Open Source Community Guidelines](https://opensource.google.com/conduct/).
|
bert-master/bert-master/LICENSE
ADDED
|
@@ -0,0 +1,202 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
|
| 2 |
+
Apache License
|
| 3 |
+
Version 2.0, January 2004
|
| 4 |
+
http://www.apache.org/licenses/
|
| 5 |
+
|
| 6 |
+
TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
|
| 7 |
+
|
| 8 |
+
1. Definitions.
|
| 9 |
+
|
| 10 |
+
"License" shall mean the terms and conditions for use, reproduction,
|
| 11 |
+
and distribution as defined by Sections 1 through 9 of this document.
|
| 12 |
+
|
| 13 |
+
"Licensor" shall mean the copyright owner or entity authorized by
|
| 14 |
+
the copyright owner that is granting the License.
|
| 15 |
+
|
| 16 |
+
"Legal Entity" shall mean the union of the acting entity and all
|
| 17 |
+
other entities that control, are controlled by, or are under common
|
| 18 |
+
control with that entity. For the purposes of this definition,
|
| 19 |
+
"control" means (i) the power, direct or indirect, to cause the
|
| 20 |
+
direction or management of such entity, whether by contract or
|
| 21 |
+
otherwise, or (ii) ownership of fifty percent (50%) or more of the
|
| 22 |
+
outstanding shares, or (iii) beneficial ownership of such entity.
|
| 23 |
+
|
| 24 |
+
"You" (or "Your") shall mean an individual or Legal Entity
|
| 25 |
+
exercising permissions granted by this License.
|
| 26 |
+
|
| 27 |
+
"Source" form shall mean the preferred form for making modifications,
|
| 28 |
+
including but not limited to software source code, documentation
|
| 29 |
+
source, and configuration files.
|
| 30 |
+
|
| 31 |
+
"Object" form shall mean any form resulting from mechanical
|
| 32 |
+
transformation or translation of a Source form, including but
|
| 33 |
+
not limited to compiled object code, generated documentation,
|
| 34 |
+
and conversions to other media types.
|
| 35 |
+
|
| 36 |
+
"Work" shall mean the work of authorship, whether in Source or
|
| 37 |
+
Object form, made available under the License, as indicated by a
|
| 38 |
+
copyright notice that is included in or attached to the work
|
| 39 |
+
(an example is provided in the Appendix below).
|
| 40 |
+
|
| 41 |
+
"Derivative Works" shall mean any work, whether in Source or Object
|
| 42 |
+
form, that is based on (or derived from) the Work and for which the
|
| 43 |
+
editorial revisions, annotations, elaborations, or other modifications
|
| 44 |
+
represent, as a whole, an original work of authorship. For the purposes
|
| 45 |
+
of this License, Derivative Works shall not include works that remain
|
| 46 |
+
separable from, or merely link (or bind by name) to the interfaces of,
|
| 47 |
+
the Work and Derivative Works thereof.
|
| 48 |
+
|
| 49 |
+
"Contribution" shall mean any work of authorship, including
|
| 50 |
+
the original version of the Work and any modifications or additions
|
| 51 |
+
to that Work or Derivative Works thereof, that is intentionally
|
| 52 |
+
submitted to Licensor for inclusion in the Work by the copyright owner
|
| 53 |
+
or by an individual or Legal Entity authorized to submit on behalf of
|
| 54 |
+
the copyright owner. For the purposes of this definition, "submitted"
|
| 55 |
+
means any form of electronic, verbal, or written communication sent
|
| 56 |
+
to the Licensor or its representatives, including but not limited to
|
| 57 |
+
communication on electronic mailing lists, source code control systems,
|
| 58 |
+
and issue tracking systems that are managed by, or on behalf of, the
|
| 59 |
+
Licensor for the purpose of discussing and improving the Work, but
|
| 60 |
+
excluding communication that is conspicuously marked or otherwise
|
| 61 |
+
designated in writing by the copyright owner as "Not a Contribution."
|
| 62 |
+
|
| 63 |
+
"Contributor" shall mean Licensor and any individual or Legal Entity
|
| 64 |
+
on behalf of whom a Contribution has been received by Licensor and
|
| 65 |
+
subsequently incorporated within the Work.
|
| 66 |
+
|
| 67 |
+
2. Grant of Copyright License. Subject to the terms and conditions of
|
| 68 |
+
this License, each Contributor hereby grants to You a perpetual,
|
| 69 |
+
worldwide, non-exclusive, no-charge, royalty-free, irrevocable
|
| 70 |
+
copyright license to reproduce, prepare Derivative Works of,
|
| 71 |
+
publicly display, publicly perform, sublicense, and distribute the
|
| 72 |
+
Work and such Derivative Works in Source or Object form.
|
| 73 |
+
|
| 74 |
+
3. Grant of Patent License. Subject to the terms and conditions of
|
| 75 |
+
this License, each Contributor hereby grants to You a perpetual,
|
| 76 |
+
worldwide, non-exclusive, no-charge, royalty-free, irrevocable
|
| 77 |
+
(except as stated in this section) patent license to make, have made,
|
| 78 |
+
use, offer to sell, sell, import, and otherwise transfer the Work,
|
| 79 |
+
where such license applies only to those patent claims licensable
|
| 80 |
+
by such Contributor that are necessarily infringed by their
|
| 81 |
+
Contribution(s) alone or by combination of their Contribution(s)
|
| 82 |
+
with the Work to which such Contribution(s) was submitted. If You
|
| 83 |
+
institute patent litigation against any entity (including a
|
| 84 |
+
cross-claim or counterclaim in a lawsuit) alleging that the Work
|
| 85 |
+
or a Contribution incorporated within the Work constitutes direct
|
| 86 |
+
or contributory patent infringement, then any patent licenses
|
| 87 |
+
granted to You under this License for that Work shall terminate
|
| 88 |
+
as of the date such litigation is filed.
|
| 89 |
+
|
| 90 |
+
4. Redistribution. You may reproduce and distribute copies of the
|
| 91 |
+
Work or Derivative Works thereof in any medium, with or without
|
| 92 |
+
modifications, and in Source or Object form, provided that You
|
| 93 |
+
meet the following conditions:
|
| 94 |
+
|
| 95 |
+
(a) You must give any other recipients of the Work or
|
| 96 |
+
Derivative Works a copy of this License; and
|
| 97 |
+
|
| 98 |
+
(b) You must cause any modified files to carry prominent notices
|
| 99 |
+
stating that You changed the files; and
|
| 100 |
+
|
| 101 |
+
(c) You must retain, in the Source form of any Derivative Works
|
| 102 |
+
that You distribute, all copyright, patent, trademark, and
|
| 103 |
+
attribution notices from the Source form of the Work,
|
| 104 |
+
excluding those notices that do not pertain to any part of
|
| 105 |
+
the Derivative Works; and
|
| 106 |
+
|
| 107 |
+
(d) If the Work includes a "NOTICE" text file as part of its
|
| 108 |
+
distribution, then any Derivative Works that You distribute must
|
| 109 |
+
include a readable copy of the attribution notices contained
|
| 110 |
+
within such NOTICE file, excluding those notices that do not
|
| 111 |
+
pertain to any part of the Derivative Works, in at least one
|
| 112 |
+
of the following places: within a NOTICE text file distributed
|
| 113 |
+
as part of the Derivative Works; within the Source form or
|
| 114 |
+
documentation, if provided along with the Derivative Works; or,
|
| 115 |
+
within a display generated by the Derivative Works, if and
|
| 116 |
+
wherever such third-party notices normally appear. The contents
|
| 117 |
+
of the NOTICE file are for informational purposes only and
|
| 118 |
+
do not modify the License. You may add Your own attribution
|
| 119 |
+
notices within Derivative Works that You distribute, alongside
|
| 120 |
+
or as an addendum to the NOTICE text from the Work, provided
|
| 121 |
+
that such additional attribution notices cannot be construed
|
| 122 |
+
as modifying the License.
|
| 123 |
+
|
| 124 |
+
You may add Your own copyright statement to Your modifications and
|
| 125 |
+
may provide additional or different license terms and conditions
|
| 126 |
+
for use, reproduction, or distribution of Your modifications, or
|
| 127 |
+
for any such Derivative Works as a whole, provided Your use,
|
| 128 |
+
reproduction, and distribution of the Work otherwise complies with
|
| 129 |
+
the conditions stated in this License.
|
| 130 |
+
|
| 131 |
+
5. Submission of Contributions. Unless You explicitly state otherwise,
|
| 132 |
+
any Contribution intentionally submitted for inclusion in the Work
|
| 133 |
+
by You to the Licensor shall be under the terms and conditions of
|
| 134 |
+
this License, without any additional terms or conditions.
|
| 135 |
+
Notwithstanding the above, nothing herein shall supersede or modify
|
| 136 |
+
the terms of any separate license agreement you may have executed
|
| 137 |
+
with Licensor regarding such Contributions.
|
| 138 |
+
|
| 139 |
+
6. Trademarks. This License does not grant permission to use the trade
|
| 140 |
+
names, trademarks, service marks, or product names of the Licensor,
|
| 141 |
+
except as required for reasonable and customary use in describing the
|
| 142 |
+
origin of the Work and reproducing the content of the NOTICE file.
|
| 143 |
+
|
| 144 |
+
7. Disclaimer of Warranty. Unless required by applicable law or
|
| 145 |
+
agreed to in writing, Licensor provides the Work (and each
|
| 146 |
+
Contributor provides its Contributions) on an "AS IS" BASIS,
|
| 147 |
+
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
|
| 148 |
+
implied, including, without limitation, any warranties or conditions
|
| 149 |
+
of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
|
| 150 |
+
PARTICULAR PURPOSE. You are solely responsible for determining the
|
| 151 |
+
appropriateness of using or redistributing the Work and assume any
|
| 152 |
+
risks associated with Your exercise of permissions under this License.
|
| 153 |
+
|
| 154 |
+
8. Limitation of Liability. In no event and under no legal theory,
|
| 155 |
+
whether in tort (including negligence), contract, or otherwise,
|
| 156 |
+
unless required by applicable law (such as deliberate and grossly
|
| 157 |
+
negligent acts) or agreed to in writing, shall any Contributor be
|
| 158 |
+
liable to You for damages, including any direct, indirect, special,
|
| 159 |
+
incidental, or consequential damages of any character arising as a
|
| 160 |
+
result of this License or out of the use or inability to use the
|
| 161 |
+
Work (including but not limited to damages for loss of goodwill,
|
| 162 |
+
work stoppage, computer failure or malfunction, or any and all
|
| 163 |
+
other commercial damages or losses), even if such Contributor
|
| 164 |
+
has been advised of the possibility of such damages.
|
| 165 |
+
|
| 166 |
+
9. Accepting Warranty or Additional Liability. While redistributing
|
| 167 |
+
the Work or Derivative Works thereof, You may choose to offer,
|
| 168 |
+
and charge a fee for, acceptance of support, warranty, indemnity,
|
| 169 |
+
or other liability obligations and/or rights consistent with this
|
| 170 |
+
License. However, in accepting such obligations, You may act only
|
| 171 |
+
on Your own behalf and on Your sole responsibility, not on behalf
|
| 172 |
+
of any other Contributor, and only if You agree to indemnify,
|
| 173 |
+
defend, and hold each Contributor harmless for any liability
|
| 174 |
+
incurred by, or claims asserted against, such Contributor by reason
|
| 175 |
+
of your accepting any such warranty or additional liability.
|
| 176 |
+
|
| 177 |
+
END OF TERMS AND CONDITIONS
|
| 178 |
+
|
| 179 |
+
APPENDIX: How to apply the Apache License to your work.
|
| 180 |
+
|
| 181 |
+
To apply the Apache License to your work, attach the following
|
| 182 |
+
boilerplate notice, with the fields enclosed by brackets "[]"
|
| 183 |
+
replaced with your own identifying information. (Don't include
|
| 184 |
+
the brackets!) The text should be enclosed in the appropriate
|
| 185 |
+
comment syntax for the file format. We also recommend that a
|
| 186 |
+
file or class name and description of purpose be included on the
|
| 187 |
+
same "printed page" as the copyright notice for easier
|
| 188 |
+
identification within third-party archives.
|
| 189 |
+
|
| 190 |
+
Copyright [yyyy] [name of copyright owner]
|
| 191 |
+
|
| 192 |
+
Licensed under the Apache License, Version 2.0 (the "License");
|
| 193 |
+
you may not use this file except in compliance with the License.
|
| 194 |
+
You may obtain a copy of the License at
|
| 195 |
+
|
| 196 |
+
http://www.apache.org/licenses/LICENSE-2.0
|
| 197 |
+
|
| 198 |
+
Unless required by applicable law or agreed to in writing, software
|
| 199 |
+
distributed under the License is distributed on an "AS IS" BASIS,
|
| 200 |
+
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
| 201 |
+
See the License for the specific language governing permissions and
|
| 202 |
+
limitations under the License.
|
bert-master/bert-master/README.md
ADDED
|
@@ -0,0 +1,1117 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# BERT
|
| 2 |
+
|
| 3 |
+
**\*\*\*\*\* New March 11th, 2020: Smaller BERT Models \*\*\*\*\***
|
| 4 |
+
|
| 5 |
+
This is a release of 24 smaller BERT models (English only, uncased, trained with WordPiece masking) referenced in [Well-Read Students Learn Better: On the Importance of Pre-training Compact Models](https://arxiv.org/abs/1908.08962).
|
| 6 |
+
|
| 7 |
+
We have shown that the standard BERT recipe (including model architecture and training objective) is effective on a wide range of model sizes, beyond BERT-Base and BERT-Large. The smaller BERT models are intended for environments with restricted computational resources. They can be fine-tuned in the same manner as the original BERT models. However, they are most effective in the context of knowledge distillation, where the fine-tuning labels are produced by a larger and more accurate teacher.
|
| 8 |
+
|
| 9 |
+
Our goal is to enable research in institutions with fewer computational resources and encourage the community to seek directions of innovation alternative to increasing model capacity.
|
| 10 |
+
|
| 11 |
+
You can download all 24 from [here][all], or individually from the table below:
|
| 12 |
+
|
| 13 |
+
| |H=128|H=256|H=512|H=768|
|
| 14 |
+
|---|:---:|:---:|:---:|:---:|
|
| 15 |
+
| **L=2** |[**2/128 (BERT-Tiny)**][2_128]|[2/256][2_256]|[2/512][2_512]|[2/768][2_768]|
|
| 16 |
+
| **L=4** |[4/128][4_128]|[**4/256 (BERT-Mini)**][4_256]|[**4/512 (BERT-Small)**][4_512]|[4/768][4_768]|
|
| 17 |
+
| **L=6** |[6/128][6_128]|[6/256][6_256]|[6/512][6_512]|[6/768][6_768]|
|
| 18 |
+
| **L=8** |[8/128][8_128]|[8/256][8_256]|[**8/512 (BERT-Medium)**][8_512]|[8/768][8_768]|
|
| 19 |
+
| **L=10** |[10/128][10_128]|[10/256][10_256]|[10/512][10_512]|[10/768][10_768]|
|
| 20 |
+
| **L=12** |[12/128][12_128]|[12/256][12_256]|[12/512][12_512]|[**12/768 (BERT-Base)**][12_768]|
|
| 21 |
+
|
| 22 |
+
Note that the BERT-Base model in this release is included for completeness only; it was re-trained under the same regime as the original model.
|
| 23 |
+
|
| 24 |
+
Here are the corresponding GLUE scores on the test set:
|
| 25 |
+
|
| 26 |
+
|Model|Score|CoLA|SST-2|MRPC|STS-B|QQP|MNLI-m|MNLI-mm|QNLI(v2)|RTE|WNLI|AX|
|
| 27 |
+
|---|:---:|:---:|:---:|:---:|:---:|:---:|:---:|:---:|:---:|:---:|:---:|:---:|
|
| 28 |
+
|BERT-Tiny|64.2|0.0|83.2|81.1/71.1|74.3/73.6|62.2/83.4|70.2|70.3|81.5|57.2|62.3|21.0|
|
| 29 |
+
|BERT-Mini|65.8|0.0|85.9|81.1/71.8|75.4/73.3|66.4/86.2|74.8|74.3|84.1|57.9|62.3|26.1|
|
| 30 |
+
|BERT-Small|71.2|27.8|89.7|83.4/76.2|78.8/77.0|68.1/87.0|77.6|77.0|86.4|61.8|62.3|28.6|
|
| 31 |
+
|BERT-Medium|73.5|38.0|89.6|86.6/81.6|80.4/78.4|69.6/87.9|80.0|79.1|87.7|62.2|62.3|30.5|
|
| 32 |
+
|
| 33 |
+
For each task, we selected the best fine-tuning hyperparameters from the lists below, and trained for 4 epochs:
|
| 34 |
+
- batch sizes: 8, 16, 32, 64, 128
|
| 35 |
+
- learning rates: 3e-4, 1e-4, 5e-5, 3e-5
|
| 36 |
+
|
| 37 |
+
If you use these models, please cite the following paper:
|
| 38 |
+
|
| 39 |
+
```
|
| 40 |
+
@article{turc2019,
|
| 41 |
+
title={Well-Read Students Learn Better: On the Importance of Pre-training Compact Models},
|
| 42 |
+
author={Turc, Iulia and Chang, Ming-Wei and Lee, Kenton and Toutanova, Kristina},
|
| 43 |
+
journal={arXiv preprint arXiv:1908.08962v2 },
|
| 44 |
+
year={2019}
|
| 45 |
+
}
|
| 46 |
+
```
|
| 47 |
+
|
| 48 |
+
[2_128]: https://storage.googleapis.com/bert_models/2020_02_20/uncased_L-2_H-128_A-2.zip
|
| 49 |
+
[2_256]: https://storage.googleapis.com/bert_models/2020_02_20/uncased_L-2_H-256_A-4.zip
|
| 50 |
+
[2_512]: https://storage.googleapis.com/bert_models/2020_02_20/uncased_L-2_H-512_A-8.zip
|
| 51 |
+
[2_768]: https://storage.googleapis.com/bert_models/2020_02_20/uncased_L-2_H-768_A-12.zip
|
| 52 |
+
[4_128]: https://storage.googleapis.com/bert_models/2020_02_20/uncased_L-4_H-128_A-2.zip
|
| 53 |
+
[4_256]: https://storage.googleapis.com/bert_models/2020_02_20/uncased_L-4_H-256_A-4.zip
|
| 54 |
+
[4_512]: https://storage.googleapis.com/bert_models/2020_02_20/uncased_L-4_H-512_A-8.zip
|
| 55 |
+
[4_768]: https://storage.googleapis.com/bert_models/2020_02_20/uncased_L-4_H-768_A-12.zip
|
| 56 |
+
[6_128]: https://storage.googleapis.com/bert_models/2020_02_20/uncased_L-6_H-128_A-2.zip
|
| 57 |
+
[6_256]: https://storage.googleapis.com/bert_models/2020_02_20/uncased_L-6_H-256_A-4.zip
|
| 58 |
+
[6_512]: https://storage.googleapis.com/bert_models/2020_02_20/uncased_L-6_H-512_A-8.zip
|
| 59 |
+
[6_768]: https://storage.googleapis.com/bert_models/2020_02_20/uncased_L-6_H-768_A-12.zip
|
| 60 |
+
[8_128]: https://storage.googleapis.com/bert_models/2020_02_20/uncased_L-8_H-128_A-2.zip
|
| 61 |
+
[8_256]: https://storage.googleapis.com/bert_models/2020_02_20/uncased_L-8_H-256_A-4.zip
|
| 62 |
+
[8_512]: https://storage.googleapis.com/bert_models/2020_02_20/uncased_L-8_H-512_A-8.zip
|
| 63 |
+
[8_768]: https://storage.googleapis.com/bert_models/2020_02_20/uncased_L-8_H-768_A-12.zip
|
| 64 |
+
[10_128]: https://storage.googleapis.com/bert_models/2020_02_20/uncased_L-10_H-128_A-2.zip
|
| 65 |
+
[10_256]: https://storage.googleapis.com/bert_models/2020_02_20/uncased_L-10_H-256_A-4.zip
|
| 66 |
+
[10_512]: https://storage.googleapis.com/bert_models/2020_02_20/uncased_L-10_H-512_A-8.zip
|
| 67 |
+
[10_768]: https://storage.googleapis.com/bert_models/2020_02_20/uncased_L-10_H-768_A-12.zip
|
| 68 |
+
[12_128]: https://storage.googleapis.com/bert_models/2020_02_20/uncased_L-12_H-128_A-2.zip
|
| 69 |
+
[12_256]: https://storage.googleapis.com/bert_models/2020_02_20/uncased_L-12_H-256_A-4.zip
|
| 70 |
+
[12_512]: https://storage.googleapis.com/bert_models/2020_02_20/uncased_L-12_H-512_A-8.zip
|
| 71 |
+
[12_768]: https://storage.googleapis.com/bert_models/2020_02_20/uncased_L-12_H-768_A-12.zip
|
| 72 |
+
[all]: https://storage.googleapis.com/bert_models/2020_02_20/all_bert_models.zip
|
| 73 |
+
|
| 74 |
+
**\*\*\*\*\* New May 31st, 2019: Whole Word Masking Models \*\*\*\*\***
|
| 75 |
+
|
| 76 |
+
This is a release of several new models which were the result of an improvement
|
| 77 |
+
the pre-processing code.
|
| 78 |
+
|
| 79 |
+
In the original pre-processing code, we randomly select WordPiece tokens to
|
| 80 |
+
mask. For example:
|
| 81 |
+
|
| 82 |
+
`Input Text: the man jumped up , put his basket on phil ##am ##mon ' s head`
|
| 83 |
+
`Original Masked Input: [MASK] man [MASK] up , put his [MASK] on phil
|
| 84 |
+
[MASK] ##mon ' s head`
|
| 85 |
+
|
| 86 |
+
The new technique is called Whole Word Masking. In this case, we always mask
|
| 87 |
+
*all* of the the tokens corresponding to a word at once. The overall masking
|
| 88 |
+
rate remains the same.
|
| 89 |
+
|
| 90 |
+
`Whole Word Masked Input: the man [MASK] up , put his basket on [MASK] [MASK]
|
| 91 |
+
[MASK] ' s head`
|
| 92 |
+
|
| 93 |
+
The training is identical -- we still predict each masked WordPiece token
|
| 94 |
+
independently. The improvement comes from the fact that the original prediction
|
| 95 |
+
task was too 'easy' for words that had been split into multiple WordPieces.
|
| 96 |
+
|
| 97 |
+
This can be enabled during data generation by passing the flag
|
| 98 |
+
`--do_whole_word_mask=True` to `create_pretraining_data.py`.
|
| 99 |
+
|
| 100 |
+
Pre-trained models with Whole Word Masking are linked below. The data and
|
| 101 |
+
training were otherwise identical, and the models have identical structure and
|
| 102 |
+
vocab to the original models. We only include BERT-Large models. When using
|
| 103 |
+
these models, please make it clear in the paper that you are using the Whole
|
| 104 |
+
Word Masking variant of BERT-Large.
|
| 105 |
+
|
| 106 |
+
* **[`BERT-Large, Uncased (Whole Word Masking)`](https://storage.googleapis.com/bert_models/2019_05_30/wwm_uncased_L-24_H-1024_A-16.zip)**:
|
| 107 |
+
24-layer, 1024-hidden, 16-heads, 340M parameters
|
| 108 |
+
|
| 109 |
+
* **[`BERT-Large, Cased (Whole Word Masking)`](https://storage.googleapis.com/bert_models/2019_05_30/wwm_cased_L-24_H-1024_A-16.zip)**:
|
| 110 |
+
24-layer, 1024-hidden, 16-heads, 340M parameters
|
| 111 |
+
|
| 112 |
+
Model | SQUAD 1.1 F1/EM | Multi NLI Accuracy
|
| 113 |
+
---------------------------------------- | :-------------: | :----------------:
|
| 114 |
+
BERT-Large, Uncased (Original) | 91.0/84.3 | 86.05
|
| 115 |
+
BERT-Large, Uncased (Whole Word Masking) | 92.8/86.7 | 87.07
|
| 116 |
+
BERT-Large, Cased (Original) | 91.5/84.8 | 86.09
|
| 117 |
+
BERT-Large, Cased (Whole Word Masking) | 92.9/86.7 | 86.46
|
| 118 |
+
|
| 119 |
+
**\*\*\*\*\* New February 7th, 2019: TfHub Module \*\*\*\*\***
|
| 120 |
+
|
| 121 |
+
BERT has been uploaded to [TensorFlow Hub](https://tfhub.dev). See
|
| 122 |
+
`run_classifier_with_tfhub.py` for an example of how to use the TF Hub module,
|
| 123 |
+
or run an example in the browser on
|
| 124 |
+
[Colab](https://colab.sandbox.google.com/github/google-research/bert/blob/master/predicting_movie_reviews_with_bert_on_tf_hub.ipynb).
|
| 125 |
+
|
| 126 |
+
**\*\*\*\*\* New November 23rd, 2018: Un-normalized multilingual model + Thai +
|
| 127 |
+
Mongolian \*\*\*\*\***
|
| 128 |
+
|
| 129 |
+
We uploaded a new multilingual model which does *not* perform any normalization
|
| 130 |
+
on the input (no lower casing, accent stripping, or Unicode normalization), and
|
| 131 |
+
additionally inclues Thai and Mongolian.
|
| 132 |
+
|
| 133 |
+
**It is recommended to use this version for developing multilingual models,
|
| 134 |
+
especially on languages with non-Latin alphabets.**
|
| 135 |
+
|
| 136 |
+
This does not require any code changes, and can be downloaded here:
|
| 137 |
+
|
| 138 |
+
* **[`BERT-Base, Multilingual Cased`](https://storage.googleapis.com/bert_models/2018_11_23/multi_cased_L-12_H-768_A-12.zip)**:
|
| 139 |
+
104 languages, 12-layer, 768-hidden, 12-heads, 110M parameters
|
| 140 |
+
|
| 141 |
+
**\*\*\*\*\* New November 15th, 2018: SOTA SQuAD 2.0 System \*\*\*\*\***
|
| 142 |
+
|
| 143 |
+
We released code changes to reproduce our 83% F1 SQuAD 2.0 system, which is
|
| 144 |
+
currently 1st place on the leaderboard by 3%. See the SQuAD 2.0 section of the
|
| 145 |
+
README for details.
|
| 146 |
+
|
| 147 |
+
**\*\*\*\*\* New November 5th, 2018: Third-party PyTorch and Chainer versions of
|
| 148 |
+
BERT available \*\*\*\*\***
|
| 149 |
+
|
| 150 |
+
NLP researchers from HuggingFace made a
|
| 151 |
+
[PyTorch version of BERT available](https://github.com/huggingface/pytorch-pretrained-BERT)
|
| 152 |
+
which is compatible with our pre-trained checkpoints and is able to reproduce
|
| 153 |
+
our results. Sosuke Kobayashi also made a
|
| 154 |
+
[Chainer version of BERT available](https://github.com/soskek/bert-chainer)
|
| 155 |
+
(Thanks!) We were not involved in the creation or maintenance of the PyTorch
|
| 156 |
+
implementation so please direct any questions towards the authors of that
|
| 157 |
+
repository.
|
| 158 |
+
|
| 159 |
+
**\*\*\*\*\* New November 3rd, 2018: Multilingual and Chinese models available
|
| 160 |
+
\*\*\*\*\***
|
| 161 |
+
|
| 162 |
+
We have made two new BERT models available:
|
| 163 |
+
|
| 164 |
+
* **[`BERT-Base, Multilingual`](https://storage.googleapis.com/bert_models/2018_11_03/multilingual_L-12_H-768_A-12.zip)
|
| 165 |
+
(Not recommended, use `Multilingual Cased` instead)**: 102 languages,
|
| 166 |
+
12-layer, 768-hidden, 12-heads, 110M parameters
|
| 167 |
+
* **[`BERT-Base, Chinese`](https://storage.googleapis.com/bert_models/2018_11_03/chinese_L-12_H-768_A-12.zip)**:
|
| 168 |
+
Chinese Simplified and Traditional, 12-layer, 768-hidden, 12-heads, 110M
|
| 169 |
+
parameters
|
| 170 |
+
|
| 171 |
+
We use character-based tokenization for Chinese, and WordPiece tokenization for
|
| 172 |
+
all other languages. Both models should work out-of-the-box without any code
|
| 173 |
+
changes. We did update the implementation of `BasicTokenizer` in
|
| 174 |
+
`tokenization.py` to support Chinese character tokenization, so please update if
|
| 175 |
+
you forked it. However, we did not change the tokenization API.
|
| 176 |
+
|
| 177 |
+
For more, see the
|
| 178 |
+
[Multilingual README](https://github.com/google-research/bert/blob/master/multilingual.md).
|
| 179 |
+
|
| 180 |
+
**\*\*\*\*\* End new information \*\*\*\*\***
|
| 181 |
+
|
| 182 |
+
## Introduction
|
| 183 |
+
|
| 184 |
+
**BERT**, or **B**idirectional **E**ncoder **R**epresentations from
|
| 185 |
+
**T**ransformers, is a new method of pre-training language representations which
|
| 186 |
+
obtains state-of-the-art results on a wide array of Natural Language Processing
|
| 187 |
+
(NLP) tasks.
|
| 188 |
+
|
| 189 |
+
Our academic paper which describes BERT in detail and provides full results on a
|
| 190 |
+
number of tasks can be found here:
|
| 191 |
+
[https://arxiv.org/abs/1810.04805](https://arxiv.org/abs/1810.04805).
|
| 192 |
+
|
| 193 |
+
To give a few numbers, here are the results on the
|
| 194 |
+
[SQuAD v1.1](https://rajpurkar.github.io/SQuAD-explorer/) question answering
|
| 195 |
+
task:
|
| 196 |
+
|
| 197 |
+
SQuAD v1.1 Leaderboard (Oct 8th 2018) | Test EM | Test F1
|
| 198 |
+
------------------------------------- | :------: | :------:
|
| 199 |
+
1st Place Ensemble - BERT | **87.4** | **93.2**
|
| 200 |
+
2nd Place Ensemble - nlnet | 86.0 | 91.7
|
| 201 |
+
1st Place Single Model - BERT | **85.1** | **91.8**
|
| 202 |
+
2nd Place Single Model - nlnet | 83.5 | 90.1
|
| 203 |
+
|
| 204 |
+
And several natural language inference tasks:
|
| 205 |
+
|
| 206 |
+
System | MultiNLI | Question NLI | SWAG
|
| 207 |
+
----------------------- | :------: | :----------: | :------:
|
| 208 |
+
BERT | **86.7** | **91.1** | **86.3**
|
| 209 |
+
OpenAI GPT (Prev. SOTA) | 82.2 | 88.1 | 75.0
|
| 210 |
+
|
| 211 |
+
Plus many other tasks.
|
| 212 |
+
|
| 213 |
+
Moreover, these results were all obtained with almost no task-specific neural
|
| 214 |
+
network architecture design.
|
| 215 |
+
|
| 216 |
+
If you already know what BERT is and you just want to get started, you can
|
| 217 |
+
[download the pre-trained models](#pre-trained-models) and
|
| 218 |
+
[run a state-of-the-art fine-tuning](#fine-tuning-with-bert) in only a few
|
| 219 |
+
minutes.
|
| 220 |
+
|
| 221 |
+
## What is BERT?
|
| 222 |
+
|
| 223 |
+
BERT is a method of pre-training language representations, meaning that we train
|
| 224 |
+
a general-purpose "language understanding" model on a large text corpus (like
|
| 225 |
+
Wikipedia), and then use that model for downstream NLP tasks that we care about
|
| 226 |
+
(like question answering). BERT outperforms previous methods because it is the
|
| 227 |
+
first *unsupervised*, *deeply bidirectional* system for pre-training NLP.
|
| 228 |
+
|
| 229 |
+
*Unsupervised* means that BERT was trained using only a plain text corpus, which
|
| 230 |
+
is important because an enormous amount of plain text data is publicly available
|
| 231 |
+
on the web in many languages.
|
| 232 |
+
|
| 233 |
+
Pre-trained representations can also either be *context-free* or *contextual*,
|
| 234 |
+
and contextual representations can further be *unidirectional* or
|
| 235 |
+
*bidirectional*. Context-free models such as
|
| 236 |
+
[word2vec](https://www.tensorflow.org/tutorials/representation/word2vec) or
|
| 237 |
+
[GloVe](https://nlp.stanford.edu/projects/glove/) generate a single "word
|
| 238 |
+
embedding" representation for each word in the vocabulary, so `bank` would have
|
| 239 |
+
the same representation in `bank deposit` and `river bank`. Contextual models
|
| 240 |
+
instead generate a representation of each word that is based on the other words
|
| 241 |
+
in the sentence.
|
| 242 |
+
|
| 243 |
+
BERT was built upon recent work in pre-training contextual representations —
|
| 244 |
+
including [Semi-supervised Sequence Learning](https://arxiv.org/abs/1511.01432),
|
| 245 |
+
[Generative Pre-Training](https://blog.openai.com/language-unsupervised/),
|
| 246 |
+
[ELMo](https://allennlp.org/elmo), and
|
| 247 |
+
[ULMFit](http://nlp.fast.ai/classification/2018/05/15/introducting-ulmfit.html)
|
| 248 |
+
— but crucially these models are all *unidirectional* or *shallowly
|
| 249 |
+
bidirectional*. This means that each word is only contextualized using the words
|
| 250 |
+
to its left (or right). For example, in the sentence `I made a bank deposit` the
|
| 251 |
+
unidirectional representation of `bank` is only based on `I made a` but not
|
| 252 |
+
`deposit`. Some previous work does combine the representations from separate
|
| 253 |
+
left-context and right-context models, but only in a "shallow" manner. BERT
|
| 254 |
+
represents "bank" using both its left and right context — `I made a ... deposit`
|
| 255 |
+
— starting from the very bottom of a deep neural network, so it is *deeply
|
| 256 |
+
bidirectional*.
|
| 257 |
+
|
| 258 |
+
BERT uses a simple approach for this: We mask out 15% of the words in the input,
|
| 259 |
+
run the entire sequence through a deep bidirectional
|
| 260 |
+
[Transformer](https://arxiv.org/abs/1706.03762) encoder, and then predict only
|
| 261 |
+
the masked words. For example:
|
| 262 |
+
|
| 263 |
+
```
|
| 264 |
+
Input: the man went to the [MASK1] . he bought a [MASK2] of milk.
|
| 265 |
+
Labels: [MASK1] = store; [MASK2] = gallon
|
| 266 |
+
```
|
| 267 |
+
|
| 268 |
+
In order to learn relationships between sentences, we also train on a simple
|
| 269 |
+
task which can be generated from any monolingual corpus: Given two sentences `A`
|
| 270 |
+
and `B`, is `B` the actual next sentence that comes after `A`, or just a random
|
| 271 |
+
sentence from the corpus?
|
| 272 |
+
|
| 273 |
+
```
|
| 274 |
+
Sentence A: the man went to the store .
|
| 275 |
+
Sentence B: he bought a gallon of milk .
|
| 276 |
+
Label: IsNextSentence
|
| 277 |
+
```
|
| 278 |
+
|
| 279 |
+
```
|
| 280 |
+
Sentence A: the man went to the store .
|
| 281 |
+
Sentence B: penguins are flightless .
|
| 282 |
+
Label: NotNextSentence
|
| 283 |
+
```
|
| 284 |
+
|
| 285 |
+
We then train a large model (12-layer to 24-layer Transformer) on a large corpus
|
| 286 |
+
(Wikipedia + [BookCorpus](http://yknzhu.wixsite.com/mbweb)) for a long time (1M
|
| 287 |
+
update steps), and that's BERT.
|
| 288 |
+
|
| 289 |
+
Using BERT has two stages: *Pre-training* and *fine-tuning*.
|
| 290 |
+
|
| 291 |
+
**Pre-training** is fairly expensive (four days on 4 to 16 Cloud TPUs), but is a
|
| 292 |
+
one-time procedure for each language (current models are English-only, but
|
| 293 |
+
multilingual models will be released in the near future). We are releasing a
|
| 294 |
+
number of pre-trained models from the paper which were pre-trained at Google.
|
| 295 |
+
Most NLP researchers will never need to pre-train their own model from scratch.
|
| 296 |
+
|
| 297 |
+
**Fine-tuning** is inexpensive. All of the results in the paper can be
|
| 298 |
+
replicated in at most 1 hour on a single Cloud TPU, or a few hours on a GPU,
|
| 299 |
+
starting from the exact same pre-trained model. SQuAD, for example, can be
|
| 300 |
+
trained in around 30 minutes on a single Cloud TPU to achieve a Dev F1 score of
|
| 301 |
+
91.0%, which is the single system state-of-the-art.
|
| 302 |
+
|
| 303 |
+
The other important aspect of BERT is that it can be adapted to many types of
|
| 304 |
+
NLP tasks very easily. In the paper, we demonstrate state-of-the-art results on
|
| 305 |
+
sentence-level (e.g., SST-2), sentence-pair-level (e.g., MultiNLI), word-level
|
| 306 |
+
(e.g., NER), and span-level (e.g., SQuAD) tasks with almost no task-specific
|
| 307 |
+
modifications.
|
| 308 |
+
|
| 309 |
+
## What has been released in this repository?
|
| 310 |
+
|
| 311 |
+
We are releasing the following:
|
| 312 |
+
|
| 313 |
+
* TensorFlow code for the BERT model architecture (which is mostly a standard
|
| 314 |
+
[Transformer](https://arxiv.org/abs/1706.03762) architecture).
|
| 315 |
+
* Pre-trained checkpoints for both the lowercase and cased version of
|
| 316 |
+
`BERT-Base` and `BERT-Large` from the paper.
|
| 317 |
+
* TensorFlow code for push-button replication of the most important
|
| 318 |
+
fine-tuning experiments from the paper, including SQuAD, MultiNLI, and MRPC.
|
| 319 |
+
|
| 320 |
+
All of the code in this repository works out-of-the-box with CPU, GPU, and Cloud
|
| 321 |
+
TPU.
|
| 322 |
+
|
| 323 |
+
## Pre-trained models
|
| 324 |
+
|
| 325 |
+
We are releasing the `BERT-Base` and `BERT-Large` models from the paper.
|
| 326 |
+
`Uncased` means that the text has been lowercased before WordPiece tokenization,
|
| 327 |
+
e.g., `John Smith` becomes `john smith`. The `Uncased` model also strips out any
|
| 328 |
+
accent markers. `Cased` means that the true case and accent markers are
|
| 329 |
+
preserved. Typically, the `Uncased` model is better unless you know that case
|
| 330 |
+
information is important for your task (e.g., Named Entity Recognition or
|
| 331 |
+
Part-of-Speech tagging).
|
| 332 |
+
|
| 333 |
+
These models are all released under the same license as the source code (Apache
|
| 334 |
+
2.0).
|
| 335 |
+
|
| 336 |
+
For information about the Multilingual and Chinese model, see the
|
| 337 |
+
[Multilingual README](https://github.com/google-research/bert/blob/master/multilingual.md).
|
| 338 |
+
|
| 339 |
+
**When using a cased model, make sure to pass `--do_lower=False` to the training
|
| 340 |
+
scripts. (Or pass `do_lower_case=False` directly to `FullTokenizer` if you're
|
| 341 |
+
using your own script.)**
|
| 342 |
+
|
| 343 |
+
The links to the models are here (right-click, 'Save link as...' on the name):
|
| 344 |
+
|
| 345 |
+
* **[`BERT-Large, Uncased (Whole Word Masking)`](https://storage.googleapis.com/bert_models/2019_05_30/wwm_uncased_L-24_H-1024_A-16.zip)**:
|
| 346 |
+
24-layer, 1024-hidden, 16-heads, 340M parameters
|
| 347 |
+
* **[`BERT-Large, Cased (Whole Word Masking)`](https://storage.googleapis.com/bert_models/2019_05_30/wwm_cased_L-24_H-1024_A-16.zip)**:
|
| 348 |
+
24-layer, 1024-hidden, 16-heads, 340M parameters
|
| 349 |
+
* **[`BERT-Base, Uncased`](https://storage.googleapis.com/bert_models/2018_10_18/uncased_L-12_H-768_A-12.zip)**:
|
| 350 |
+
12-layer, 768-hidden, 12-heads, 110M parameters
|
| 351 |
+
* **[`BERT-Large, Uncased`](https://storage.googleapis.com/bert_models/2018_10_18/uncased_L-24_H-1024_A-16.zip)**:
|
| 352 |
+
24-layer, 1024-hidden, 16-heads, 340M parameters
|
| 353 |
+
* **[`BERT-Base, Cased`](https://storage.googleapis.com/bert_models/2018_10_18/cased_L-12_H-768_A-12.zip)**:
|
| 354 |
+
12-layer, 768-hidden, 12-heads , 110M parameters
|
| 355 |
+
* **[`BERT-Large, Cased`](https://storage.googleapis.com/bert_models/2018_10_18/cased_L-24_H-1024_A-16.zip)**:
|
| 356 |
+
24-layer, 1024-hidden, 16-heads, 340M parameters
|
| 357 |
+
* **[`BERT-Base, Multilingual Cased (New, recommended)`](https://storage.googleapis.com/bert_models/2018_11_23/multi_cased_L-12_H-768_A-12.zip)**:
|
| 358 |
+
104 languages, 12-layer, 768-hidden, 12-heads, 110M parameters
|
| 359 |
+
* **[`BERT-Base, Multilingual Uncased (Orig, not recommended)`](https://storage.googleapis.com/bert_models/2018_11_03/multilingual_L-12_H-768_A-12.zip)
|
| 360 |
+
(Not recommended, use `Multilingual Cased` instead)**: 102 languages,
|
| 361 |
+
12-layer, 768-hidden, 12-heads, 110M parameters
|
| 362 |
+
* **[`BERT-Base, Chinese`](https://storage.googleapis.com/bert_models/2018_11_03/chinese_L-12_H-768_A-12.zip)**:
|
| 363 |
+
Chinese Simplified and Traditional, 12-layer, 768-hidden, 12-heads, 110M
|
| 364 |
+
parameters
|
| 365 |
+
|
| 366 |
+
Each .zip file contains three items:
|
| 367 |
+
|
| 368 |
+
* A TensorFlow checkpoint (`bert_model.ckpt`) containing the pre-trained
|
| 369 |
+
weights (which is actually 3 files).
|
| 370 |
+
* A vocab file (`vocab.txt`) to map WordPiece to word id.
|
| 371 |
+
* A config file (`bert_config.json`) which specifies the hyperparameters of
|
| 372 |
+
the model.
|
| 373 |
+
|
| 374 |
+
## Fine-tuning with BERT
|
| 375 |
+
|
| 376 |
+
**Important**: All results on the paper were fine-tuned on a single Cloud TPU,
|
| 377 |
+
which has 64GB of RAM. It is currently not possible to re-produce most of the
|
| 378 |
+
`BERT-Large` results on the paper using a GPU with 12GB - 16GB of RAM, because
|
| 379 |
+
the maximum batch size that can fit in memory is too small. We are working on
|
| 380 |
+
adding code to this repository which allows for much larger effective batch size
|
| 381 |
+
on the GPU. See the section on [out-of-memory issues](#out-of-memory-issues) for
|
| 382 |
+
more details.
|
| 383 |
+
|
| 384 |
+
This code was tested with TensorFlow 1.11.0. It was tested with Python2 and
|
| 385 |
+
Python3 (but more thoroughly with Python2, since this is what's used internally
|
| 386 |
+
in Google).
|
| 387 |
+
|
| 388 |
+
The fine-tuning examples which use `BERT-Base` should be able to run on a GPU
|
| 389 |
+
that has at least 12GB of RAM using the hyperparameters given.
|
| 390 |
+
|
| 391 |
+
### Fine-tuning with Cloud TPUs
|
| 392 |
+
|
| 393 |
+
Most of the examples below assumes that you will be running training/evaluation
|
| 394 |
+
on your local machine, using a GPU like a Titan X or GTX 1080.
|
| 395 |
+
|
| 396 |
+
However, if you have access to a Cloud TPU that you want to train on, just add
|
| 397 |
+
the following flags to `run_classifier.py` or `run_squad.py`:
|
| 398 |
+
|
| 399 |
+
```
|
| 400 |
+
--use_tpu=True \
|
| 401 |
+
--tpu_name=$TPU_NAME
|
| 402 |
+
```
|
| 403 |
+
|
| 404 |
+
Please see the
|
| 405 |
+
[Google Cloud TPU tutorial](https://cloud.google.com/tpu/docs/tutorials/mnist)
|
| 406 |
+
for how to use Cloud TPUs. Alternatively, you can use the Google Colab notebook
|
| 407 |
+
"[BERT FineTuning with Cloud TPUs](https://colab.research.google.com/github/tensorflow/tpu/blob/master/tools/colab/bert_finetuning_with_cloud_tpus.ipynb)".
|
| 408 |
+
|
| 409 |
+
On Cloud TPUs, the pretrained model and the output directory will need to be on
|
| 410 |
+
Google Cloud Storage. For example, if you have a bucket named `some_bucket`, you
|
| 411 |
+
might use the following flags instead:
|
| 412 |
+
|
| 413 |
+
```
|
| 414 |
+
--output_dir=gs://some_bucket/my_output_dir/
|
| 415 |
+
```
|
| 416 |
+
|
| 417 |
+
The unzipped pre-trained model files can also be found in the Google Cloud
|
| 418 |
+
Storage folder `gs://bert_models/2018_10_18`. For example:
|
| 419 |
+
|
| 420 |
+
```
|
| 421 |
+
export BERT_BASE_DIR=gs://bert_models/2018_10_18/uncased_L-12_H-768_A-12
|
| 422 |
+
```
|
| 423 |
+
|
| 424 |
+
### Sentence (and sentence-pair) classification tasks
|
| 425 |
+
|
| 426 |
+
Before running this example you must download the
|
| 427 |
+
[GLUE data](https://gluebenchmark.com/tasks) by running
|
| 428 |
+
[this script](https://gist.github.com/W4ngatang/60c2bdb54d156a41194446737ce03e2e)
|
| 429 |
+
and unpack it to some directory `$GLUE_DIR`. Next, download the `BERT-Base`
|
| 430 |
+
checkpoint and unzip it to some directory `$BERT_BASE_DIR`.
|
| 431 |
+
|
| 432 |
+
This example code fine-tunes `BERT-Base` on the Microsoft Research Paraphrase
|
| 433 |
+
Corpus (MRPC) corpus, which only contains 3,600 examples and can fine-tune in a
|
| 434 |
+
few minutes on most GPUs.
|
| 435 |
+
|
| 436 |
+
```shell
|
| 437 |
+
export BERT_BASE_DIR=/path/to/bert/uncased_L-12_H-768_A-12
|
| 438 |
+
export GLUE_DIR=/path/to/glue
|
| 439 |
+
|
| 440 |
+
python run_classifier.py \
|
| 441 |
+
--task_name=MRPC \
|
| 442 |
+
--do_train=true \
|
| 443 |
+
--do_eval=true \
|
| 444 |
+
--data_dir=$GLUE_DIR/MRPC \
|
| 445 |
+
--vocab_file=$BERT_BASE_DIR/vocab.txt \
|
| 446 |
+
--bert_config_file=$BERT_BASE_DIR/bert_config.json \
|
| 447 |
+
--init_checkpoint=$BERT_BASE_DIR/bert_model.ckpt \
|
| 448 |
+
--max_seq_length=128 \
|
| 449 |
+
--train_batch_size=32 \
|
| 450 |
+
--learning_rate=2e-5 \
|
| 451 |
+
--num_train_epochs=3.0 \
|
| 452 |
+
--output_dir=/tmp/mrpc_output/
|
| 453 |
+
```
|
| 454 |
+
|
| 455 |
+
You should see output like this:
|
| 456 |
+
|
| 457 |
+
```
|
| 458 |
+
***** Eval results *****
|
| 459 |
+
eval_accuracy = 0.845588
|
| 460 |
+
eval_loss = 0.505248
|
| 461 |
+
global_step = 343
|
| 462 |
+
loss = 0.505248
|
| 463 |
+
```
|
| 464 |
+
|
| 465 |
+
This means that the Dev set accuracy was 84.55%. Small sets like MRPC have a
|
| 466 |
+
high variance in the Dev set accuracy, even when starting from the same
|
| 467 |
+
pre-training checkpoint. If you re-run multiple times (making sure to point to
|
| 468 |
+
different `output_dir`), you should see results between 84% and 88%.
|
| 469 |
+
|
| 470 |
+
A few other pre-trained models are implemented off-the-shelf in
|
| 471 |
+
`run_classifier.py`, so it should be straightforward to follow those examples to
|
| 472 |
+
use BERT for any single-sentence or sentence-pair classification task.
|
| 473 |
+
|
| 474 |
+
Note: You might see a message `Running train on CPU`. This really just means
|
| 475 |
+
that it's running on something other than a Cloud TPU, which includes a GPU.
|
| 476 |
+
|
| 477 |
+
#### Prediction from classifier
|
| 478 |
+
|
| 479 |
+
Once you have trained your classifier you can use it in inference mode by using
|
| 480 |
+
the --do_predict=true command. You need to have a file named test.tsv in the
|
| 481 |
+
input folder. Output will be created in file called test_results.tsv in the
|
| 482 |
+
output folder. Each line will contain output for each sample, columns are the
|
| 483 |
+
class probabilities.
|
| 484 |
+
|
| 485 |
+
```shell
|
| 486 |
+
export BERT_BASE_DIR=/path/to/bert/uncased_L-12_H-768_A-12
|
| 487 |
+
export GLUE_DIR=/path/to/glue
|
| 488 |
+
export TRAINED_CLASSIFIER=/path/to/fine/tuned/classifier
|
| 489 |
+
|
| 490 |
+
python run_classifier.py \
|
| 491 |
+
--task_name=MRPC \
|
| 492 |
+
--do_predict=true \
|
| 493 |
+
--data_dir=$GLUE_DIR/MRPC \
|
| 494 |
+
--vocab_file=$BERT_BASE_DIR/vocab.txt \
|
| 495 |
+
--bert_config_file=$BERT_BASE_DIR/bert_config.json \
|
| 496 |
+
--init_checkpoint=$TRAINED_CLASSIFIER \
|
| 497 |
+
--max_seq_length=128 \
|
| 498 |
+
--output_dir=/tmp/mrpc_output/
|
| 499 |
+
```
|
| 500 |
+
|
| 501 |
+
### SQuAD 1.1
|
| 502 |
+
|
| 503 |
+
The Stanford Question Answering Dataset (SQuAD) is a popular question answering
|
| 504 |
+
benchmark dataset. BERT (at the time of the release) obtains state-of-the-art
|
| 505 |
+
results on SQuAD with almost no task-specific network architecture modifications
|
| 506 |
+
or data augmentation. However, it does require semi-complex data pre-processing
|
| 507 |
+
and post-processing to deal with (a) the variable-length nature of SQuAD context
|
| 508 |
+
paragraphs, and (b) the character-level answer annotations which are used for
|
| 509 |
+
SQuAD training. This processing is implemented and documented in `run_squad.py`.
|
| 510 |
+
|
| 511 |
+
To run on SQuAD, you will first need to download the dataset. The
|
| 512 |
+
[SQuAD website](https://rajpurkar.github.io/SQuAD-explorer/) does not seem to
|
| 513 |
+
link to the v1.1 datasets any longer, but the necessary files can be found here:
|
| 514 |
+
|
| 515 |
+
* [train-v1.1.json](https://rajpurkar.github.io/SQuAD-explorer/dataset/train-v1.1.json)
|
| 516 |
+
* [dev-v1.1.json](https://rajpurkar.github.io/SQuAD-explorer/dataset/dev-v1.1.json)
|
| 517 |
+
* [evaluate-v1.1.py](https://github.com/allenai/bi-att-flow/blob/master/squad/evaluate-v1.1.py)
|
| 518 |
+
|
| 519 |
+
Download these to some directory `$SQUAD_DIR`.
|
| 520 |
+
|
| 521 |
+
The state-of-the-art SQuAD results from the paper currently cannot be reproduced
|
| 522 |
+
on a 12GB-16GB GPU due to memory constraints (in fact, even batch size 1 does
|
| 523 |
+
not seem to fit on a 12GB GPU using `BERT-Large`). However, a reasonably strong
|
| 524 |
+
`BERT-Base` model can be trained on the GPU with these hyperparameters:
|
| 525 |
+
|
| 526 |
+
```shell
|
| 527 |
+
python run_squad.py \
|
| 528 |
+
--vocab_file=$BERT_BASE_DIR/vocab.txt \
|
| 529 |
+
--bert_config_file=$BERT_BASE_DIR/bert_config.json \
|
| 530 |
+
--init_checkpoint=$BERT_BASE_DIR/bert_model.ckpt \
|
| 531 |
+
--do_train=True \
|
| 532 |
+
--train_file=$SQUAD_DIR/train-v1.1.json \
|
| 533 |
+
--do_predict=True \
|
| 534 |
+
--predict_file=$SQUAD_DIR/dev-v1.1.json \
|
| 535 |
+
--train_batch_size=12 \
|
| 536 |
+
--learning_rate=3e-5 \
|
| 537 |
+
--num_train_epochs=2.0 \
|
| 538 |
+
--max_seq_length=384 \
|
| 539 |
+
--doc_stride=128 \
|
| 540 |
+
--output_dir=/tmp/squad_base/
|
| 541 |
+
```
|
| 542 |
+
|
| 543 |
+
The dev set predictions will be saved into a file called `predictions.json` in
|
| 544 |
+
the `output_dir`:
|
| 545 |
+
|
| 546 |
+
```shell
|
| 547 |
+
python $SQUAD_DIR/evaluate-v1.1.py $SQUAD_DIR/dev-v1.1.json ./squad/predictions.json
|
| 548 |
+
```
|
| 549 |
+
|
| 550 |
+
Which should produce an output like this:
|
| 551 |
+
|
| 552 |
+
```shell
|
| 553 |
+
{"f1": 88.41249612335034, "exact_match": 81.2488174077578}
|
| 554 |
+
```
|
| 555 |
+
|
| 556 |
+
You should see a result similar to the 88.5% reported in the paper for
|
| 557 |
+
`BERT-Base`.
|
| 558 |
+
|
| 559 |
+
If you have access to a Cloud TPU, you can train with `BERT-Large`. Here is a
|
| 560 |
+
set of hyperparameters (slightly different than the paper) which consistently
|
| 561 |
+
obtain around 90.5%-91.0% F1 single-system trained only on SQuAD:
|
| 562 |
+
|
| 563 |
+
```shell
|
| 564 |
+
python run_squad.py \
|
| 565 |
+
--vocab_file=$BERT_LARGE_DIR/vocab.txt \
|
| 566 |
+
--bert_config_file=$BERT_LARGE_DIR/bert_config.json \
|
| 567 |
+
--init_checkpoint=$BERT_LARGE_DIR/bert_model.ckpt \
|
| 568 |
+
--do_train=True \
|
| 569 |
+
--train_file=$SQUAD_DIR/train-v1.1.json \
|
| 570 |
+
--do_predict=True \
|
| 571 |
+
--predict_file=$SQUAD_DIR/dev-v1.1.json \
|
| 572 |
+
--train_batch_size=24 \
|
| 573 |
+
--learning_rate=3e-5 \
|
| 574 |
+
--num_train_epochs=2.0 \
|
| 575 |
+
--max_seq_length=384 \
|
| 576 |
+
--doc_stride=128 \
|
| 577 |
+
--output_dir=gs://some_bucket/squad_large/ \
|
| 578 |
+
--use_tpu=True \
|
| 579 |
+
--tpu_name=$TPU_NAME
|
| 580 |
+
```
|
| 581 |
+
|
| 582 |
+
For example, one random run with these parameters produces the following Dev
|
| 583 |
+
scores:
|
| 584 |
+
|
| 585 |
+
```shell
|
| 586 |
+
{"f1": 90.87081895814865, "exact_match": 84.38978240302744}
|
| 587 |
+
```
|
| 588 |
+
|
| 589 |
+
If you fine-tune for one epoch on
|
| 590 |
+
[TriviaQA](http://nlp.cs.washington.edu/triviaqa/) before this the results will
|
| 591 |
+
be even better, but you will need to convert TriviaQA into the SQuAD json
|
| 592 |
+
format.
|
| 593 |
+
|
| 594 |
+
### SQuAD 2.0
|
| 595 |
+
|
| 596 |
+
This model is also implemented and documented in `run_squad.py`.
|
| 597 |
+
|
| 598 |
+
To run on SQuAD 2.0, you will first need to download the dataset. The necessary
|
| 599 |
+
files can be found here:
|
| 600 |
+
|
| 601 |
+
* [train-v2.0.json](https://rajpurkar.github.io/SQuAD-explorer/dataset/train-v2.0.json)
|
| 602 |
+
* [dev-v2.0.json](https://rajpurkar.github.io/SQuAD-explorer/dataset/dev-v2.0.json)
|
| 603 |
+
* [evaluate-v2.0.py](https://worksheets.codalab.org/rest/bundles/0x6b567e1cf2e041ec80d7098f031c5c9e/contents/blob/)
|
| 604 |
+
|
| 605 |
+
Download these to some directory `$SQUAD_DIR`.
|
| 606 |
+
|
| 607 |
+
On Cloud TPU you can run with BERT-Large as follows:
|
| 608 |
+
|
| 609 |
+
```shell
|
| 610 |
+
python run_squad.py \
|
| 611 |
+
--vocab_file=$BERT_LARGE_DIR/vocab.txt \
|
| 612 |
+
--bert_config_file=$BERT_LARGE_DIR/bert_config.json \
|
| 613 |
+
--init_checkpoint=$BERT_LARGE_DIR/bert_model.ckpt \
|
| 614 |
+
--do_train=True \
|
| 615 |
+
--train_file=$SQUAD_DIR/train-v2.0.json \
|
| 616 |
+
--do_predict=True \
|
| 617 |
+
--predict_file=$SQUAD_DIR/dev-v2.0.json \
|
| 618 |
+
--train_batch_size=24 \
|
| 619 |
+
--learning_rate=3e-5 \
|
| 620 |
+
--num_train_epochs=2.0 \
|
| 621 |
+
--max_seq_length=384 \
|
| 622 |
+
--doc_stride=128 \
|
| 623 |
+
--output_dir=gs://some_bucket/squad_large/ \
|
| 624 |
+
--use_tpu=True \
|
| 625 |
+
--tpu_name=$TPU_NAME \
|
| 626 |
+
--version_2_with_negative=True
|
| 627 |
+
```
|
| 628 |
+
|
| 629 |
+
We assume you have copied everything from the output directory to a local
|
| 630 |
+
directory called ./squad/. The initial dev set predictions will be at
|
| 631 |
+
./squad/predictions.json and the differences between the score of no answer ("")
|
| 632 |
+
and the best non-null answer for each question will be in the file
|
| 633 |
+
./squad/null_odds.json
|
| 634 |
+
|
| 635 |
+
Run this script to tune a threshold for predicting null versus non-null answers:
|
| 636 |
+
|
| 637 |
+
python $SQUAD_DIR/evaluate-v2.0.py $SQUAD_DIR/dev-v2.0.json
|
| 638 |
+
./squad/predictions.json --na-prob-file ./squad/null_odds.json
|
| 639 |
+
|
| 640 |
+
Assume the script outputs "best_f1_thresh" THRESH. (Typical values are between
|
| 641 |
+
-1.0 and -5.0). You can now re-run the model to generate predictions with the
|
| 642 |
+
derived threshold or alternatively you can extract the appropriate answers from
|
| 643 |
+
./squad/nbest_predictions.json.
|
| 644 |
+
|
| 645 |
+
```shell
|
| 646 |
+
python run_squad.py \
|
| 647 |
+
--vocab_file=$BERT_LARGE_DIR/vocab.txt \
|
| 648 |
+
--bert_config_file=$BERT_LARGE_DIR/bert_config.json \
|
| 649 |
+
--init_checkpoint=$BERT_LARGE_DIR/bert_model.ckpt \
|
| 650 |
+
--do_train=False \
|
| 651 |
+
--train_file=$SQUAD_DIR/train-v2.0.json \
|
| 652 |
+
--do_predict=True \
|
| 653 |
+
--predict_file=$SQUAD_DIR/dev-v2.0.json \
|
| 654 |
+
--train_batch_size=24 \
|
| 655 |
+
--learning_rate=3e-5 \
|
| 656 |
+
--num_train_epochs=2.0 \
|
| 657 |
+
--max_seq_length=384 \
|
| 658 |
+
--doc_stride=128 \
|
| 659 |
+
--output_dir=gs://some_bucket/squad_large/ \
|
| 660 |
+
--use_tpu=True \
|
| 661 |
+
--tpu_name=$TPU_NAME \
|
| 662 |
+
--version_2_with_negative=True \
|
| 663 |
+
--null_score_diff_threshold=$THRESH
|
| 664 |
+
```
|
| 665 |
+
|
| 666 |
+
### Out-of-memory issues
|
| 667 |
+
|
| 668 |
+
All experiments in the paper were fine-tuned on a Cloud TPU, which has 64GB of
|
| 669 |
+
device RAM. Therefore, when using a GPU with 12GB - 16GB of RAM, you are likely
|
| 670 |
+
to encounter out-of-memory issues if you use the same hyperparameters described
|
| 671 |
+
in the paper.
|
| 672 |
+
|
| 673 |
+
The factors that affect memory usage are:
|
| 674 |
+
|
| 675 |
+
* **`max_seq_length`**: The released models were trained with sequence lengths
|
| 676 |
+
up to 512, but you can fine-tune with a shorter max sequence length to save
|
| 677 |
+
substantial memory. This is controlled by the `max_seq_length` flag in our
|
| 678 |
+
example code.
|
| 679 |
+
|
| 680 |
+
* **`train_batch_size`**: The memory usage is also directly proportional to
|
| 681 |
+
the batch size.
|
| 682 |
+
|
| 683 |
+
* **Model type, `BERT-Base` vs. `BERT-Large`**: The `BERT-Large` model
|
| 684 |
+
requires significantly more memory than `BERT-Base`.
|
| 685 |
+
|
| 686 |
+
* **Optimizer**: The default optimizer for BERT is Adam, which requires a lot
|
| 687 |
+
of extra memory to store the `m` and `v` vectors. Switching to a more memory
|
| 688 |
+
efficient optimizer can reduce memory usage, but can also affect the
|
| 689 |
+
results. We have not experimented with other optimizers for fine-tuning.
|
| 690 |
+
|
| 691 |
+
Using the default training scripts (`run_classifier.py` and `run_squad.py`), we
|
| 692 |
+
benchmarked the maximum batch size on single Titan X GPU (12GB RAM) with
|
| 693 |
+
TensorFlow 1.11.0:
|
| 694 |
+
|
| 695 |
+
System | Seq Length | Max Batch Size
|
| 696 |
+
------------ | ---------- | --------------
|
| 697 |
+
`BERT-Base` | 64 | 64
|
| 698 |
+
... | 128 | 32
|
| 699 |
+
... | 256 | 16
|
| 700 |
+
... | 320 | 14
|
| 701 |
+
... | 384 | 12
|
| 702 |
+
... | 512 | 6
|
| 703 |
+
`BERT-Large` | 64 | 12
|
| 704 |
+
... | 128 | 6
|
| 705 |
+
... | 256 | 2
|
| 706 |
+
... | 320 | 1
|
| 707 |
+
... | 384 | 0
|
| 708 |
+
... | 512 | 0
|
| 709 |
+
|
| 710 |
+
Unfortunately, these max batch sizes for `BERT-Large` are so small that they
|
| 711 |
+
will actually harm the model accuracy, regardless of the learning rate used. We
|
| 712 |
+
are working on adding code to this repository which will allow much larger
|
| 713 |
+
effective batch sizes to be used on the GPU. The code will be based on one (or
|
| 714 |
+
both) of the following techniques:
|
| 715 |
+
|
| 716 |
+
* **Gradient accumulation**: The samples in a minibatch are typically
|
| 717 |
+
independent with respect to gradient computation (excluding batch
|
| 718 |
+
normalization, which is not used here). This means that the gradients of
|
| 719 |
+
multiple smaller minibatches can be accumulated before performing the weight
|
| 720 |
+
update, and this will be exactly equivalent to a single larger update.
|
| 721 |
+
|
| 722 |
+
* [**Gradient checkpointing**](https://github.com/openai/gradient-checkpointing):
|
| 723 |
+
The major use of GPU/TPU memory during DNN training is caching the
|
| 724 |
+
intermediate activations in the forward pass that are necessary for
|
| 725 |
+
efficient computation in the backward pass. "Gradient checkpointing" trades
|
| 726 |
+
memory for compute time by re-computing the activations in an intelligent
|
| 727 |
+
way.
|
| 728 |
+
|
| 729 |
+
**However, this is not implemented in the current release.**
|
| 730 |
+
|
| 731 |
+
## Using BERT to extract fixed feature vectors (like ELMo)
|
| 732 |
+
|
| 733 |
+
In certain cases, rather than fine-tuning the entire pre-trained model
|
| 734 |
+
end-to-end, it can be beneficial to obtained *pre-trained contextual
|
| 735 |
+
embeddings*, which are fixed contextual representations of each input token
|
| 736 |
+
generated from the hidden layers of the pre-trained model. This should also
|
| 737 |
+
mitigate most of the out-of-memory issues.
|
| 738 |
+
|
| 739 |
+
As an example, we include the script `extract_features.py` which can be used
|
| 740 |
+
like this:
|
| 741 |
+
|
| 742 |
+
```shell
|
| 743 |
+
# Sentence A and Sentence B are separated by the ||| delimiter for sentence
|
| 744 |
+
# pair tasks like question answering and entailment.
|
| 745 |
+
# For single sentence inputs, put one sentence per line and DON'T use the
|
| 746 |
+
# delimiter.
|
| 747 |
+
echo 'Who was Jim Henson ? ||| Jim Henson was a puppeteer' > /tmp/input.txt
|
| 748 |
+
|
| 749 |
+
python extract_features.py \
|
| 750 |
+
--input_file=/tmp/input.txt \
|
| 751 |
+
--output_file=/tmp/output.jsonl \
|
| 752 |
+
--vocab_file=$BERT_BASE_DIR/vocab.txt \
|
| 753 |
+
--bert_config_file=$BERT_BASE_DIR/bert_config.json \
|
| 754 |
+
--init_checkpoint=$BERT_BASE_DIR/bert_model.ckpt \
|
| 755 |
+
--layers=-1,-2,-3,-4 \
|
| 756 |
+
--max_seq_length=128 \
|
| 757 |
+
--batch_size=8
|
| 758 |
+
```
|
| 759 |
+
|
| 760 |
+
This will create a JSON file (one line per line of input) containing the BERT
|
| 761 |
+
activations from each Transformer layer specified by `layers` (-1 is the final
|
| 762 |
+
hidden layer of the Transformer, etc.)
|
| 763 |
+
|
| 764 |
+
Note that this script will produce very large output files (by default, around
|
| 765 |
+
15kb for every input token).
|
| 766 |
+
|
| 767 |
+
If you need to maintain alignment between the original and tokenized words (for
|
| 768 |
+
projecting training labels), see the [Tokenization](#tokenization) section
|
| 769 |
+
below.
|
| 770 |
+
|
| 771 |
+
**Note:** You may see a message like `Could not find trained model in model_dir:
|
| 772 |
+
/tmp/tmpuB5g5c, running initialization to predict.` This message is expected, it
|
| 773 |
+
just means that we are using the `init_from_checkpoint()` API rather than the
|
| 774 |
+
saved model API. If you don't specify a checkpoint or specify an invalid
|
| 775 |
+
checkpoint, this script will complain.
|
| 776 |
+
|
| 777 |
+
## Tokenization
|
| 778 |
+
|
| 779 |
+
For sentence-level tasks (or sentence-pair) tasks, tokenization is very simple.
|
| 780 |
+
Just follow the example code in `run_classifier.py` and `extract_features.py`.
|
| 781 |
+
The basic procedure for sentence-level tasks is:
|
| 782 |
+
|
| 783 |
+
1. Instantiate an instance of `tokenizer = tokenization.FullTokenizer`
|
| 784 |
+
|
| 785 |
+
2. Tokenize the raw text with `tokens = tokenizer.tokenize(raw_text)`.
|
| 786 |
+
|
| 787 |
+
3. Truncate to the maximum sequence length. (You can use up to 512, but you
|
| 788 |
+
probably want to use shorter if possible for memory and speed reasons.)
|
| 789 |
+
|
| 790 |
+
4. Add the `[CLS]` and `[SEP]` tokens in the right place.
|
| 791 |
+
|
| 792 |
+
Word-level and span-level tasks (e.g., SQuAD and NER) are more complex, since
|
| 793 |
+
you need to maintain alignment between your input text and output text so that
|
| 794 |
+
you can project your training labels. SQuAD is a particularly complex example
|
| 795 |
+
because the input labels are *character*-based, and SQuAD paragraphs are often
|
| 796 |
+
longer than our maximum sequence length. See the code in `run_squad.py` to show
|
| 797 |
+
how we handle this.
|
| 798 |
+
|
| 799 |
+
Before we describe the general recipe for handling word-level tasks, it's
|
| 800 |
+
important to understand what exactly our tokenizer is doing. It has three main
|
| 801 |
+
steps:
|
| 802 |
+
|
| 803 |
+
1. **Text normalization**: Convert all whitespace characters to spaces, and
|
| 804 |
+
(for the `Uncased` model) lowercase the input and strip out accent markers.
|
| 805 |
+
E.g., `John Johanson's, → john johanson's,`.
|
| 806 |
+
|
| 807 |
+
2. **Punctuation splitting**: Split *all* punctuation characters on both sides
|
| 808 |
+
(i.e., add whitespace around all punctuation characters). Punctuation
|
| 809 |
+
characters are defined as (a) Anything with a `P*` Unicode class, (b) any
|
| 810 |
+
non-letter/number/space ASCII character (e.g., characters like `$` which are
|
| 811 |
+
technically not punctuation). E.g., `john johanson's, → john johanson ' s ,`
|
| 812 |
+
|
| 813 |
+
3. **WordPiece tokenization**: Apply whitespace tokenization to the output of
|
| 814 |
+
the above procedure, and apply
|
| 815 |
+
[WordPiece](https://github.com/tensorflow/tensor2tensor/blob/master/tensor2tensor/data_generators/text_encoder.py)
|
| 816 |
+
tokenization to each token separately. (Our implementation is directly based
|
| 817 |
+
on the one from `tensor2tensor`, which is linked). E.g., `john johanson ' s
|
| 818 |
+
, → john johan ##son ' s ,`
|
| 819 |
+
|
| 820 |
+
The advantage of this scheme is that it is "compatible" with most existing
|
| 821 |
+
English tokenizers. For example, imagine that you have a part-of-speech tagging
|
| 822 |
+
task which looks like this:
|
| 823 |
+
|
| 824 |
+
```
|
| 825 |
+
Input: John Johanson 's house
|
| 826 |
+
Labels: NNP NNP POS NN
|
| 827 |
+
```
|
| 828 |
+
|
| 829 |
+
The tokenized output will look like this:
|
| 830 |
+
|
| 831 |
+
```
|
| 832 |
+
Tokens: john johan ##son ' s house
|
| 833 |
+
```
|
| 834 |
+
|
| 835 |
+
Crucially, this would be the same output as if the raw text were `John
|
| 836 |
+
Johanson's house` (with no space before the `'s`).
|
| 837 |
+
|
| 838 |
+
If you have a pre-tokenized representation with word-level annotations, you can
|
| 839 |
+
simply tokenize each input word independently, and deterministically maintain an
|
| 840 |
+
original-to-tokenized alignment:
|
| 841 |
+
|
| 842 |
+
```python
|
| 843 |
+
### Input
|
| 844 |
+
orig_tokens = ["John", "Johanson", "'s", "house"]
|
| 845 |
+
labels = ["NNP", "NNP", "POS", "NN"]
|
| 846 |
+
|
| 847 |
+
### Output
|
| 848 |
+
bert_tokens = []
|
| 849 |
+
|
| 850 |
+
# Token map will be an int -> int mapping between the `orig_tokens` index and
|
| 851 |
+
# the `bert_tokens` index.
|
| 852 |
+
orig_to_tok_map = []
|
| 853 |
+
|
| 854 |
+
tokenizer = tokenization.FullTokenizer(
|
| 855 |
+
vocab_file=vocab_file, do_lower_case=True)
|
| 856 |
+
|
| 857 |
+
bert_tokens.append("[CLS]")
|
| 858 |
+
for orig_token in orig_tokens:
|
| 859 |
+
orig_to_tok_map.append(len(bert_tokens))
|
| 860 |
+
bert_tokens.extend(tokenizer.tokenize(orig_token))
|
| 861 |
+
bert_tokens.append("[SEP]")
|
| 862 |
+
|
| 863 |
+
# bert_tokens == ["[CLS]", "john", "johan", "##son", "'", "s", "house", "[SEP]"]
|
| 864 |
+
# orig_to_tok_map == [1, 2, 4, 6]
|
| 865 |
+
```
|
| 866 |
+
|
| 867 |
+
Now `orig_to_tok_map` can be used to project `labels` to the tokenized
|
| 868 |
+
representation.
|
| 869 |
+
|
| 870 |
+
There are common English tokenization schemes which will cause a slight mismatch
|
| 871 |
+
between how BERT was pre-trained. For example, if your input tokenization splits
|
| 872 |
+
off contractions like `do n't`, this will cause a mismatch. If it is possible to
|
| 873 |
+
do so, you should pre-process your data to convert these back to raw-looking
|
| 874 |
+
text, but if it's not possible, this mismatch is likely not a big deal.
|
| 875 |
+
|
| 876 |
+
## Pre-training with BERT
|
| 877 |
+
|
| 878 |
+
We are releasing code to do "masked LM" and "next sentence prediction" on an
|
| 879 |
+
arbitrary text corpus. Note that this is *not* the exact code that was used for
|
| 880 |
+
the paper (the original code was written in C++, and had some additional
|
| 881 |
+
complexity), but this code does generate pre-training data as described in the
|
| 882 |
+
paper.
|
| 883 |
+
|
| 884 |
+
Here's how to run the data generation. The input is a plain text file, with one
|
| 885 |
+
sentence per line. (It is important that these be actual sentences for the "next
|
| 886 |
+
sentence prediction" task). Documents are delimited by empty lines. The output
|
| 887 |
+
is a set of `tf.train.Example`s serialized into `TFRecord` file format.
|
| 888 |
+
|
| 889 |
+
You can perform sentence segmentation with an off-the-shelf NLP toolkit such as
|
| 890 |
+
[spaCy](https://spacy.io/). The `create_pretraining_data.py` script will
|
| 891 |
+
concatenate segments until they reach the maximum sequence length to minimize
|
| 892 |
+
computational waste from padding (see the script for more details). However, you
|
| 893 |
+
may want to intentionally add a slight amount of noise to your input data (e.g.,
|
| 894 |
+
randomly truncate 2% of input segments) to make it more robust to non-sentential
|
| 895 |
+
input during fine-tuning.
|
| 896 |
+
|
| 897 |
+
This script stores all of the examples for the entire input file in memory, so
|
| 898 |
+
for large data files you should shard the input file and call the script
|
| 899 |
+
multiple times. (You can pass in a file glob to `run_pretraining.py`, e.g.,
|
| 900 |
+
`tf_examples.tf_record*`.)
|
| 901 |
+
|
| 902 |
+
The `max_predictions_per_seq` is the maximum number of masked LM predictions per
|
| 903 |
+
sequence. You should set this to around `max_seq_length` * `masked_lm_prob` (the
|
| 904 |
+
script doesn't do that automatically because the exact value needs to be passed
|
| 905 |
+
to both scripts).
|
| 906 |
+
|
| 907 |
+
```shell
|
| 908 |
+
python create_pretraining_data.py \
|
| 909 |
+
--input_file=./sample_text.txt \
|
| 910 |
+
--output_file=/tmp/tf_examples.tfrecord \
|
| 911 |
+
--vocab_file=$BERT_BASE_DIR/vocab.txt \
|
| 912 |
+
--do_lower_case=True \
|
| 913 |
+
--max_seq_length=128 \
|
| 914 |
+
--max_predictions_per_seq=20 \
|
| 915 |
+
--masked_lm_prob=0.15 \
|
| 916 |
+
--random_seed=12345 \
|
| 917 |
+
--dupe_factor=5
|
| 918 |
+
```
|
| 919 |
+
|
| 920 |
+
Here's how to run the pre-training. Do not include `init_checkpoint` if you are
|
| 921 |
+
pre-training from scratch. The model configuration (including vocab size) is
|
| 922 |
+
specified in `bert_config_file`. This demo code only pre-trains for a small
|
| 923 |
+
number of steps (20), but in practice you will probably want to set
|
| 924 |
+
`num_train_steps` to 10000 steps or more. The `max_seq_length` and
|
| 925 |
+
`max_predictions_per_seq` parameters passed to `run_pretraining.py` must be the
|
| 926 |
+
same as `create_pretraining_data.py`.
|
| 927 |
+
|
| 928 |
+
```shell
|
| 929 |
+
python run_pretraining.py \
|
| 930 |
+
--input_file=/tmp/tf_examples.tfrecord \
|
| 931 |
+
--output_dir=/tmp/pretraining_output \
|
| 932 |
+
--do_train=True \
|
| 933 |
+
--do_eval=True \
|
| 934 |
+
--bert_config_file=$BERT_BASE_DIR/bert_config.json \
|
| 935 |
+
--init_checkpoint=$BERT_BASE_DIR/bert_model.ckpt \
|
| 936 |
+
--train_batch_size=32 \
|
| 937 |
+
--max_seq_length=128 \
|
| 938 |
+
--max_predictions_per_seq=20 \
|
| 939 |
+
--num_train_steps=20 \
|
| 940 |
+
--num_warmup_steps=10 \
|
| 941 |
+
--learning_rate=2e-5
|
| 942 |
+
```
|
| 943 |
+
|
| 944 |
+
This will produce an output like this:
|
| 945 |
+
|
| 946 |
+
```
|
| 947 |
+
***** Eval results *****
|
| 948 |
+
global_step = 20
|
| 949 |
+
loss = 0.0979674
|
| 950 |
+
masked_lm_accuracy = 0.985479
|
| 951 |
+
masked_lm_loss = 0.0979328
|
| 952 |
+
next_sentence_accuracy = 1.0
|
| 953 |
+
next_sentence_loss = 3.45724e-05
|
| 954 |
+
```
|
| 955 |
+
|
| 956 |
+
Note that since our `sample_text.txt` file is very small, this example training
|
| 957 |
+
will overfit that data in only a few steps and produce unrealistically high
|
| 958 |
+
accuracy numbers.
|
| 959 |
+
|
| 960 |
+
### Pre-training tips and caveats
|
| 961 |
+
|
| 962 |
+
* **If using your own vocabulary, make sure to change `vocab_size` in
|
| 963 |
+
`bert_config.json`. If you use a larger vocabulary without changing this,
|
| 964 |
+
you will likely get NaNs when training on GPU or TPU due to unchecked
|
| 965 |
+
out-of-bounds access.**
|
| 966 |
+
* If your task has a large domain-specific corpus available (e.g., "movie
|
| 967 |
+
reviews" or "scientific papers"), it will likely be beneficial to run
|
| 968 |
+
additional steps of pre-training on your corpus, starting from the BERT
|
| 969 |
+
checkpoint.
|
| 970 |
+
* The learning rate we used in the paper was 1e-4. However, if you are doing
|
| 971 |
+
additional steps of pre-training starting from an existing BERT checkpoint,
|
| 972 |
+
you should use a smaller learning rate (e.g., 2e-5).
|
| 973 |
+
* Current BERT models are English-only, but we do plan to release a
|
| 974 |
+
multilingual model which has been pre-trained on a lot of languages in the
|
| 975 |
+
near future (hopefully by the end of November 2018).
|
| 976 |
+
* Longer sequences are disproportionately expensive because attention is
|
| 977 |
+
quadratic to the sequence length. In other words, a batch of 64 sequences of
|
| 978 |
+
length 512 is much more expensive than a batch of 256 sequences of
|
| 979 |
+
length 128. The fully-connected/convolutional cost is the same, but the
|
| 980 |
+
attention cost is far greater for the 512-length sequences. Therefore, one
|
| 981 |
+
good recipe is to pre-train for, say, 90,000 steps with a sequence length of
|
| 982 |
+
128 and then for 10,000 additional steps with a sequence length of 512. The
|
| 983 |
+
very long sequences are mostly needed to learn positional embeddings, which
|
| 984 |
+
can be learned fairly quickly. Note that this does require generating the
|
| 985 |
+
data twice with different values of `max_seq_length`.
|
| 986 |
+
* If you are pre-training from scratch, be prepared that pre-training is
|
| 987 |
+
computationally expensive, especially on GPUs. If you are pre-training from
|
| 988 |
+
scratch, our recommended recipe is to pre-train a `BERT-Base` on a single
|
| 989 |
+
[preemptible Cloud TPU v2](https://cloud.google.com/tpu/docs/pricing), which
|
| 990 |
+
takes about 2 weeks at a cost of about $500 USD (based on the pricing in
|
| 991 |
+
October 2018). You will have to scale down the batch size when only training
|
| 992 |
+
on a single Cloud TPU, compared to what was used in the paper. It is
|
| 993 |
+
recommended to use the largest batch size that fits into TPU memory.
|
| 994 |
+
|
| 995 |
+
### Pre-training data
|
| 996 |
+
|
| 997 |
+
We will **not** be able to release the pre-processed datasets used in the paper.
|
| 998 |
+
For Wikipedia, the recommended pre-processing is to download
|
| 999 |
+
[the latest dump](https://dumps.wikimedia.org/enwiki/latest/enwiki-latest-pages-articles.xml.bz2),
|
| 1000 |
+
extract the text with
|
| 1001 |
+
[`WikiExtractor.py`](https://github.com/attardi/wikiextractor), and then apply
|
| 1002 |
+
any necessary cleanup to convert it into plain text.
|
| 1003 |
+
|
| 1004 |
+
Unfortunately the researchers who collected the
|
| 1005 |
+
[BookCorpus](http://yknzhu.wixsite.com/mbweb) no longer have it available for
|
| 1006 |
+
public download. The
|
| 1007 |
+
[Project Guttenberg Dataset](https://web.eecs.umich.edu/~lahiri/gutenberg_dataset.html)
|
| 1008 |
+
is a somewhat smaller (200M word) collection of older books that are public
|
| 1009 |
+
domain.
|
| 1010 |
+
|
| 1011 |
+
[Common Crawl](http://commoncrawl.org/) is another very large collection of
|
| 1012 |
+
text, but you will likely have to do substantial pre-processing and cleanup to
|
| 1013 |
+
extract a usable corpus for pre-training BERT.
|
| 1014 |
+
|
| 1015 |
+
### Learning a new WordPiece vocabulary
|
| 1016 |
+
|
| 1017 |
+
This repository does not include code for *learning* a new WordPiece vocabulary.
|
| 1018 |
+
The reason is that the code used in the paper was implemented in C++ with
|
| 1019 |
+
dependencies on Google's internal libraries. For English, it is almost always
|
| 1020 |
+
better to just start with our vocabulary and pre-trained models. For learning
|
| 1021 |
+
vocabularies of other languages, there are a number of open source options
|
| 1022 |
+
available. However, keep in mind that these are not compatible with our
|
| 1023 |
+
`tokenization.py` library:
|
| 1024 |
+
|
| 1025 |
+
* [Google's SentencePiece library](https://github.com/google/sentencepiece)
|
| 1026 |
+
|
| 1027 |
+
* [tensor2tensor's WordPiece generation script](https://github.com/tensorflow/tensor2tensor/blob/master/tensor2tensor/data_generators/text_encoder_build_subword.py)
|
| 1028 |
+
|
| 1029 |
+
* [Rico Sennrich's Byte Pair Encoding library](https://github.com/rsennrich/subword-nmt)
|
| 1030 |
+
|
| 1031 |
+
## Using BERT in Colab
|
| 1032 |
+
|
| 1033 |
+
If you want to use BERT with [Colab](https://colab.research.google.com), you can
|
| 1034 |
+
get started with the notebook
|
| 1035 |
+
"[BERT FineTuning with Cloud TPUs](https://colab.research.google.com/github/tensorflow/tpu/blob/master/tools/colab/bert_finetuning_with_cloud_tpus.ipynb)".
|
| 1036 |
+
**At the time of this writing (October 31st, 2018), Colab users can access a
|
| 1037 |
+
Cloud TPU completely for free.** Note: One per user, availability limited,
|
| 1038 |
+
requires a Google Cloud Platform account with storage (although storage may be
|
| 1039 |
+
purchased with free credit for signing up with GCP), and this capability may not
|
| 1040 |
+
longer be available in the future. Click on the BERT Colab that was just linked
|
| 1041 |
+
for more information.
|
| 1042 |
+
|
| 1043 |
+
## FAQ
|
| 1044 |
+
|
| 1045 |
+
#### Is this code compatible with Cloud TPUs? What about GPUs?
|
| 1046 |
+
|
| 1047 |
+
Yes, all of the code in this repository works out-of-the-box with CPU, GPU, and
|
| 1048 |
+
Cloud TPU. However, GPU training is single-GPU only.
|
| 1049 |
+
|
| 1050 |
+
#### I am getting out-of-memory errors, what is wrong?
|
| 1051 |
+
|
| 1052 |
+
See the section on [out-of-memory issues](#out-of-memory-issues) for more
|
| 1053 |
+
information.
|
| 1054 |
+
|
| 1055 |
+
#### Is there a PyTorch version available?
|
| 1056 |
+
|
| 1057 |
+
There is no official PyTorch implementation. However, NLP researchers from
|
| 1058 |
+
HuggingFace made a
|
| 1059 |
+
[PyTorch version of BERT available](https://github.com/huggingface/pytorch-pretrained-BERT)
|
| 1060 |
+
which is compatible with our pre-trained checkpoints and is able to reproduce
|
| 1061 |
+
our results. We were not involved in the creation or maintenance of the PyTorch
|
| 1062 |
+
implementation so please direct any questions towards the authors of that
|
| 1063 |
+
repository.
|
| 1064 |
+
|
| 1065 |
+
#### Is there a Chainer version available?
|
| 1066 |
+
|
| 1067 |
+
There is no official Chainer implementation. However, Sosuke Kobayashi made a
|
| 1068 |
+
[Chainer version of BERT available](https://github.com/soskek/bert-chainer)
|
| 1069 |
+
which is compatible with our pre-trained checkpoints and is able to reproduce
|
| 1070 |
+
our results. We were not involved in the creation or maintenance of the Chainer
|
| 1071 |
+
implementation so please direct any questions towards the authors of that
|
| 1072 |
+
repository.
|
| 1073 |
+
|
| 1074 |
+
#### Will models in other languages be released?
|
| 1075 |
+
|
| 1076 |
+
Yes, we plan to release a multi-lingual BERT model in the near future. We cannot
|
| 1077 |
+
make promises about exactly which languages will be included, but it will likely
|
| 1078 |
+
be a single model which includes *most* of the languages which have a
|
| 1079 |
+
significantly-sized Wikipedia.
|
| 1080 |
+
|
| 1081 |
+
#### Will models larger than `BERT-Large` be released?
|
| 1082 |
+
|
| 1083 |
+
So far we have not attempted to train anything larger than `BERT-Large`. It is
|
| 1084 |
+
possible that we will release larger models if we are able to obtain significant
|
| 1085 |
+
improvements.
|
| 1086 |
+
|
| 1087 |
+
#### What license is this library released under?
|
| 1088 |
+
|
| 1089 |
+
All code *and* models are released under the Apache 2.0 license. See the
|
| 1090 |
+
`LICENSE` file for more information.
|
| 1091 |
+
|
| 1092 |
+
#### How do I cite BERT?
|
| 1093 |
+
|
| 1094 |
+
For now, cite [the Arxiv paper](https://arxiv.org/abs/1810.04805):
|
| 1095 |
+
|
| 1096 |
+
```
|
| 1097 |
+
@article{devlin2018bert,
|
| 1098 |
+
title={BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding},
|
| 1099 |
+
author={Devlin, Jacob and Chang, Ming-Wei and Lee, Kenton and Toutanova, Kristina},
|
| 1100 |
+
journal={arXiv preprint arXiv:1810.04805},
|
| 1101 |
+
year={2018}
|
| 1102 |
+
}
|
| 1103 |
+
```
|
| 1104 |
+
|
| 1105 |
+
If we submit the paper to a conference or journal, we will update the BibTeX.
|
| 1106 |
+
|
| 1107 |
+
## Disclaimer
|
| 1108 |
+
|
| 1109 |
+
This is not an official Google product.
|
| 1110 |
+
|
| 1111 |
+
## Contact information
|
| 1112 |
+
|
| 1113 |
+
For help or issues using BERT, please submit a GitHub issue.
|
| 1114 |
+
|
| 1115 |
+
For personal communication related to BERT, please contact Jacob Devlin
|
| 1116 |
+
(`jacobdevlin@google.com`), Ming-Wei Chang (`mingweichang@google.com`), or
|
| 1117 |
+
Kenton Lee (`kentonl@google.com`).
|
bert-master/bert-master/__init__.py
ADDED
|
@@ -0,0 +1,15 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# coding=utf-8
|
| 2 |
+
# Copyright 2018 The Google AI Language Team Authors.
|
| 3 |
+
#
|
| 4 |
+
# Licensed under the Apache License, Version 2.0 (the "License");
|
| 5 |
+
# you may not use this file except in compliance with the License.
|
| 6 |
+
# You may obtain a copy of the License at
|
| 7 |
+
#
|
| 8 |
+
# http://www.apache.org/licenses/LICENSE-2.0
|
| 9 |
+
#
|
| 10 |
+
# Unless required by applicable law or agreed to in writing, software
|
| 11 |
+
# distributed under the License is distributed on an "AS IS" BASIS,
|
| 12 |
+
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
| 13 |
+
# See the License for the specific language governing permissions and
|
| 14 |
+
# limitations under the License.
|
| 15 |
+
|
bert-master/bert-master/create_pretraining_data.py
ADDED
|
@@ -0,0 +1,469 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# coding=utf-8
|
| 2 |
+
# Copyright 2018 The Google AI Language Team Authors.
|
| 3 |
+
#
|
| 4 |
+
# Licensed under the Apache License, Version 2.0 (the "License");
|
| 5 |
+
# you may not use this file except in compliance with the License.
|
| 6 |
+
# You may obtain a copy of the License at
|
| 7 |
+
#
|
| 8 |
+
# http://www.apache.org/licenses/LICENSE-2.0
|
| 9 |
+
#
|
| 10 |
+
# Unless required by applicable law or agreed to in writing, software
|
| 11 |
+
# distributed under the License is distributed on an "AS IS" BASIS,
|
| 12 |
+
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
| 13 |
+
# See the License for the specific language governing permissions and
|
| 14 |
+
# limitations under the License.
|
| 15 |
+
"""Create masked LM/next sentence masked_lm TF examples for BERT."""
|
| 16 |
+
|
| 17 |
+
from __future__ import absolute_import
|
| 18 |
+
from __future__ import division
|
| 19 |
+
from __future__ import print_function
|
| 20 |
+
|
| 21 |
+
import collections
|
| 22 |
+
import random
|
| 23 |
+
import tokenization
|
| 24 |
+
import tensorflow as tf
|
| 25 |
+
|
| 26 |
+
flags = tf.flags
|
| 27 |
+
|
| 28 |
+
FLAGS = flags.FLAGS
|
| 29 |
+
|
| 30 |
+
flags.DEFINE_string("input_file", None,
|
| 31 |
+
"Input raw text file (or comma-separated list of files).")
|
| 32 |
+
|
| 33 |
+
flags.DEFINE_string(
|
| 34 |
+
"output_file", None,
|
| 35 |
+
"Output TF example file (or comma-separated list of files).")
|
| 36 |
+
|
| 37 |
+
flags.DEFINE_string("vocab_file", None,
|
| 38 |
+
"The vocabulary file that the BERT model was trained on.")
|
| 39 |
+
|
| 40 |
+
flags.DEFINE_bool(
|
| 41 |
+
"do_lower_case", True,
|
| 42 |
+
"Whether to lower case the input text. Should be True for uncased "
|
| 43 |
+
"models and False for cased models.")
|
| 44 |
+
|
| 45 |
+
flags.DEFINE_bool(
|
| 46 |
+
"do_whole_word_mask", False,
|
| 47 |
+
"Whether to use whole word masking rather than per-WordPiece masking.")
|
| 48 |
+
|
| 49 |
+
flags.DEFINE_integer("max_seq_length", 128, "Maximum sequence length.")
|
| 50 |
+
|
| 51 |
+
flags.DEFINE_integer("max_predictions_per_seq", 20,
|
| 52 |
+
"Maximum number of masked LM predictions per sequence.")
|
| 53 |
+
|
| 54 |
+
flags.DEFINE_integer("random_seed", 12345, "Random seed for data generation.")
|
| 55 |
+
|
| 56 |
+
flags.DEFINE_integer(
|
| 57 |
+
"dupe_factor", 10,
|
| 58 |
+
"Number of times to duplicate the input data (with different masks).")
|
| 59 |
+
|
| 60 |
+
flags.DEFINE_float("masked_lm_prob", 0.15, "Masked LM probability.")
|
| 61 |
+
|
| 62 |
+
flags.DEFINE_float(
|
| 63 |
+
"short_seq_prob", 0.1,
|
| 64 |
+
"Probability of creating sequences which are shorter than the "
|
| 65 |
+
"maximum length.")
|
| 66 |
+
|
| 67 |
+
|
| 68 |
+
class TrainingInstance(object):
|
| 69 |
+
"""A single training instance (sentence pair)."""
|
| 70 |
+
|
| 71 |
+
def __init__(self, tokens, segment_ids, masked_lm_positions, masked_lm_labels,
|
| 72 |
+
is_random_next):
|
| 73 |
+
self.tokens = tokens
|
| 74 |
+
self.segment_ids = segment_ids
|
| 75 |
+
self.is_random_next = is_random_next
|
| 76 |
+
self.masked_lm_positions = masked_lm_positions
|
| 77 |
+
self.masked_lm_labels = masked_lm_labels
|
| 78 |
+
|
| 79 |
+
def __str__(self):
|
| 80 |
+
s = ""
|
| 81 |
+
s += "tokens: %s\n" % (" ".join(
|
| 82 |
+
[tokenization.printable_text(x) for x in self.tokens]))
|
| 83 |
+
s += "segment_ids: %s\n" % (" ".join([str(x) for x in self.segment_ids]))
|
| 84 |
+
s += "is_random_next: %s\n" % self.is_random_next
|
| 85 |
+
s += "masked_lm_positions: %s\n" % (" ".join(
|
| 86 |
+
[str(x) for x in self.masked_lm_positions]))
|
| 87 |
+
s += "masked_lm_labels: %s\n" % (" ".join(
|
| 88 |
+
[tokenization.printable_text(x) for x in self.masked_lm_labels]))
|
| 89 |
+
s += "\n"
|
| 90 |
+
return s
|
| 91 |
+
|
| 92 |
+
def __repr__(self):
|
| 93 |
+
return self.__str__()
|
| 94 |
+
|
| 95 |
+
|
| 96 |
+
def write_instance_to_example_files(instances, tokenizer, max_seq_length,
|
| 97 |
+
max_predictions_per_seq, output_files):
|
| 98 |
+
"""Create TF example files from `TrainingInstance`s."""
|
| 99 |
+
writers = []
|
| 100 |
+
for output_file in output_files:
|
| 101 |
+
writers.append(tf.python_io.TFRecordWriter(output_file))
|
| 102 |
+
|
| 103 |
+
writer_index = 0
|
| 104 |
+
|
| 105 |
+
total_written = 0
|
| 106 |
+
for (inst_index, instance) in enumerate(instances):
|
| 107 |
+
input_ids = tokenizer.convert_tokens_to_ids(instance.tokens)
|
| 108 |
+
input_mask = [1] * len(input_ids)
|
| 109 |
+
segment_ids = list(instance.segment_ids)
|
| 110 |
+
assert len(input_ids) <= max_seq_length
|
| 111 |
+
|
| 112 |
+
while len(input_ids) < max_seq_length:
|
| 113 |
+
input_ids.append(0)
|
| 114 |
+
input_mask.append(0)
|
| 115 |
+
segment_ids.append(0)
|
| 116 |
+
|
| 117 |
+
assert len(input_ids) == max_seq_length
|
| 118 |
+
assert len(input_mask) == max_seq_length
|
| 119 |
+
assert len(segment_ids) == max_seq_length
|
| 120 |
+
|
| 121 |
+
masked_lm_positions = list(instance.masked_lm_positions)
|
| 122 |
+
masked_lm_ids = tokenizer.convert_tokens_to_ids(instance.masked_lm_labels)
|
| 123 |
+
masked_lm_weights = [1.0] * len(masked_lm_ids)
|
| 124 |
+
|
| 125 |
+
while len(masked_lm_positions) < max_predictions_per_seq:
|
| 126 |
+
masked_lm_positions.append(0)
|
| 127 |
+
masked_lm_ids.append(0)
|
| 128 |
+
masked_lm_weights.append(0.0)
|
| 129 |
+
|
| 130 |
+
next_sentence_label = 1 if instance.is_random_next else 0
|
| 131 |
+
|
| 132 |
+
features = collections.OrderedDict()
|
| 133 |
+
features["input_ids"] = create_int_feature(input_ids)
|
| 134 |
+
features["input_mask"] = create_int_feature(input_mask)
|
| 135 |
+
features["segment_ids"] = create_int_feature(segment_ids)
|
| 136 |
+
features["masked_lm_positions"] = create_int_feature(masked_lm_positions)
|
| 137 |
+
features["masked_lm_ids"] = create_int_feature(masked_lm_ids)
|
| 138 |
+
features["masked_lm_weights"] = create_float_feature(masked_lm_weights)
|
| 139 |
+
features["next_sentence_labels"] = create_int_feature([next_sentence_label])
|
| 140 |
+
|
| 141 |
+
tf_example = tf.train.Example(features=tf.train.Features(feature=features))
|
| 142 |
+
|
| 143 |
+
writers[writer_index].write(tf_example.SerializeToString())
|
| 144 |
+
writer_index = (writer_index + 1) % len(writers)
|
| 145 |
+
|
| 146 |
+
total_written += 1
|
| 147 |
+
|
| 148 |
+
if inst_index < 20:
|
| 149 |
+
tf.logging.info("*** Example ***")
|
| 150 |
+
tf.logging.info("tokens: %s" % " ".join(
|
| 151 |
+
[tokenization.printable_text(x) for x in instance.tokens]))
|
| 152 |
+
|
| 153 |
+
for feature_name in features.keys():
|
| 154 |
+
feature = features[feature_name]
|
| 155 |
+
values = []
|
| 156 |
+
if feature.int64_list.value:
|
| 157 |
+
values = feature.int64_list.value
|
| 158 |
+
elif feature.float_list.value:
|
| 159 |
+
values = feature.float_list.value
|
| 160 |
+
tf.logging.info(
|
| 161 |
+
"%s: %s" % (feature_name, " ".join([str(x) for x in values])))
|
| 162 |
+
|
| 163 |
+
for writer in writers:
|
| 164 |
+
writer.close()
|
| 165 |
+
|
| 166 |
+
tf.logging.info("Wrote %d total instances", total_written)
|
| 167 |
+
|
| 168 |
+
|
| 169 |
+
def create_int_feature(values):
|
| 170 |
+
feature = tf.train.Feature(int64_list=tf.train.Int64List(value=list(values)))
|
| 171 |
+
return feature
|
| 172 |
+
|
| 173 |
+
|
| 174 |
+
def create_float_feature(values):
|
| 175 |
+
feature = tf.train.Feature(float_list=tf.train.FloatList(value=list(values)))
|
| 176 |
+
return feature
|
| 177 |
+
|
| 178 |
+
|
| 179 |
+
def create_training_instances(input_files, tokenizer, max_seq_length,
|
| 180 |
+
dupe_factor, short_seq_prob, masked_lm_prob,
|
| 181 |
+
max_predictions_per_seq, rng):
|
| 182 |
+
"""Create `TrainingInstance`s from raw text."""
|
| 183 |
+
all_documents = [[]]
|
| 184 |
+
|
| 185 |
+
# Input file format:
|
| 186 |
+
# (1) One sentence per line. These should ideally be actual sentences, not
|
| 187 |
+
# entire paragraphs or arbitrary spans of text. (Because we use the
|
| 188 |
+
# sentence boundaries for the "next sentence prediction" task).
|
| 189 |
+
# (2) Blank lines between documents. Document boundaries are needed so
|
| 190 |
+
# that the "next sentence prediction" task doesn't span between documents.
|
| 191 |
+
for input_file in input_files:
|
| 192 |
+
with tf.gfile.GFile(input_file, "r") as reader:
|
| 193 |
+
while True:
|
| 194 |
+
line = tokenization.convert_to_unicode(reader.readline())
|
| 195 |
+
if not line:
|
| 196 |
+
break
|
| 197 |
+
line = line.strip()
|
| 198 |
+
|
| 199 |
+
# Empty lines are used as document delimiters
|
| 200 |
+
if not line:
|
| 201 |
+
all_documents.append([])
|
| 202 |
+
tokens = tokenizer.tokenize(line)
|
| 203 |
+
if tokens:
|
| 204 |
+
all_documents[-1].append(tokens)
|
| 205 |
+
|
| 206 |
+
# Remove empty documents
|
| 207 |
+
all_documents = [x for x in all_documents if x]
|
| 208 |
+
rng.shuffle(all_documents)
|
| 209 |
+
|
| 210 |
+
vocab_words = list(tokenizer.vocab.keys())
|
| 211 |
+
instances = []
|
| 212 |
+
for _ in range(dupe_factor):
|
| 213 |
+
for document_index in range(len(all_documents)):
|
| 214 |
+
instances.extend(
|
| 215 |
+
create_instances_from_document(
|
| 216 |
+
all_documents, document_index, max_seq_length, short_seq_prob,
|
| 217 |
+
masked_lm_prob, max_predictions_per_seq, vocab_words, rng))
|
| 218 |
+
|
| 219 |
+
rng.shuffle(instances)
|
| 220 |
+
return instances
|
| 221 |
+
|
| 222 |
+
|
| 223 |
+
def create_instances_from_document(
|
| 224 |
+
all_documents, document_index, max_seq_length, short_seq_prob,
|
| 225 |
+
masked_lm_prob, max_predictions_per_seq, vocab_words, rng):
|
| 226 |
+
"""Creates `TrainingInstance`s for a single document."""
|
| 227 |
+
document = all_documents[document_index]
|
| 228 |
+
|
| 229 |
+
# Account for [CLS], [SEP], [SEP]
|
| 230 |
+
max_num_tokens = max_seq_length - 3
|
| 231 |
+
|
| 232 |
+
# We *usually* want to fill up the entire sequence since we are padding
|
| 233 |
+
# to `max_seq_length` anyways, so short sequences are generally wasted
|
| 234 |
+
# computation. However, we *sometimes*
|
| 235 |
+
# (i.e., short_seq_prob == 0.1 == 10% of the time) want to use shorter
|
| 236 |
+
# sequences to minimize the mismatch between pre-training and fine-tuning.
|
| 237 |
+
# The `target_seq_length` is just a rough target however, whereas
|
| 238 |
+
# `max_seq_length` is a hard limit.
|
| 239 |
+
target_seq_length = max_num_tokens
|
| 240 |
+
if rng.random() < short_seq_prob:
|
| 241 |
+
target_seq_length = rng.randint(2, max_num_tokens)
|
| 242 |
+
|
| 243 |
+
# We DON'T just concatenate all of the tokens from a document into a long
|
| 244 |
+
# sequence and choose an arbitrary split point because this would make the
|
| 245 |
+
# next sentence prediction task too easy. Instead, we split the input into
|
| 246 |
+
# segments "A" and "B" based on the actual "sentences" provided by the user
|
| 247 |
+
# input.
|
| 248 |
+
instances = []
|
| 249 |
+
current_chunk = []
|
| 250 |
+
current_length = 0
|
| 251 |
+
i = 0
|
| 252 |
+
while i < len(document):
|
| 253 |
+
segment = document[i]
|
| 254 |
+
current_chunk.append(segment)
|
| 255 |
+
current_length += len(segment)
|
| 256 |
+
if i == len(document) - 1 or current_length >= target_seq_length:
|
| 257 |
+
if current_chunk:
|
| 258 |
+
# `a_end` is how many segments from `current_chunk` go into the `A`
|
| 259 |
+
# (first) sentence.
|
| 260 |
+
a_end = 1
|
| 261 |
+
if len(current_chunk) >= 2:
|
| 262 |
+
a_end = rng.randint(1, len(current_chunk) - 1)
|
| 263 |
+
|
| 264 |
+
tokens_a = []
|
| 265 |
+
for j in range(a_end):
|
| 266 |
+
tokens_a.extend(current_chunk[j])
|
| 267 |
+
|
| 268 |
+
tokens_b = []
|
| 269 |
+
# Random next
|
| 270 |
+
is_random_next = False
|
| 271 |
+
if len(current_chunk) == 1 or rng.random() < 0.5:
|
| 272 |
+
is_random_next = True
|
| 273 |
+
target_b_length = target_seq_length - len(tokens_a)
|
| 274 |
+
|
| 275 |
+
# This should rarely go for more than one iteration for large
|
| 276 |
+
# corpora. However, just to be careful, we try to make sure that
|
| 277 |
+
# the random document is not the same as the document
|
| 278 |
+
# we're processing.
|
| 279 |
+
for _ in range(10):
|
| 280 |
+
random_document_index = rng.randint(0, len(all_documents) - 1)
|
| 281 |
+
if random_document_index != document_index:
|
| 282 |
+
break
|
| 283 |
+
|
| 284 |
+
random_document = all_documents[random_document_index]
|
| 285 |
+
random_start = rng.randint(0, len(random_document) - 1)
|
| 286 |
+
for j in range(random_start, len(random_document)):
|
| 287 |
+
tokens_b.extend(random_document[j])
|
| 288 |
+
if len(tokens_b) >= target_b_length:
|
| 289 |
+
break
|
| 290 |
+
# We didn't actually use these segments so we "put them back" so
|
| 291 |
+
# they don't go to waste.
|
| 292 |
+
num_unused_segments = len(current_chunk) - a_end
|
| 293 |
+
i -= num_unused_segments
|
| 294 |
+
# Actual next
|
| 295 |
+
else:
|
| 296 |
+
is_random_next = False
|
| 297 |
+
for j in range(a_end, len(current_chunk)):
|
| 298 |
+
tokens_b.extend(current_chunk[j])
|
| 299 |
+
truncate_seq_pair(tokens_a, tokens_b, max_num_tokens, rng)
|
| 300 |
+
|
| 301 |
+
assert len(tokens_a) >= 1
|
| 302 |
+
assert len(tokens_b) >= 1
|
| 303 |
+
|
| 304 |
+
tokens = []
|
| 305 |
+
segment_ids = []
|
| 306 |
+
tokens.append("[CLS]")
|
| 307 |
+
segment_ids.append(0)
|
| 308 |
+
for token in tokens_a:
|
| 309 |
+
tokens.append(token)
|
| 310 |
+
segment_ids.append(0)
|
| 311 |
+
|
| 312 |
+
tokens.append("[SEP]")
|
| 313 |
+
segment_ids.append(0)
|
| 314 |
+
|
| 315 |
+
for token in tokens_b:
|
| 316 |
+
tokens.append(token)
|
| 317 |
+
segment_ids.append(1)
|
| 318 |
+
tokens.append("[SEP]")
|
| 319 |
+
segment_ids.append(1)
|
| 320 |
+
|
| 321 |
+
(tokens, masked_lm_positions,
|
| 322 |
+
masked_lm_labels) = create_masked_lm_predictions(
|
| 323 |
+
tokens, masked_lm_prob, max_predictions_per_seq, vocab_words, rng)
|
| 324 |
+
instance = TrainingInstance(
|
| 325 |
+
tokens=tokens,
|
| 326 |
+
segment_ids=segment_ids,
|
| 327 |
+
is_random_next=is_random_next,
|
| 328 |
+
masked_lm_positions=masked_lm_positions,
|
| 329 |
+
masked_lm_labels=masked_lm_labels)
|
| 330 |
+
instances.append(instance)
|
| 331 |
+
current_chunk = []
|
| 332 |
+
current_length = 0
|
| 333 |
+
i += 1
|
| 334 |
+
|
| 335 |
+
return instances
|
| 336 |
+
|
| 337 |
+
|
| 338 |
+
MaskedLmInstance = collections.namedtuple("MaskedLmInstance",
|
| 339 |
+
["index", "label"])
|
| 340 |
+
|
| 341 |
+
|
| 342 |
+
def create_masked_lm_predictions(tokens, masked_lm_prob,
|
| 343 |
+
max_predictions_per_seq, vocab_words, rng):
|
| 344 |
+
"""Creates the predictions for the masked LM objective."""
|
| 345 |
+
|
| 346 |
+
cand_indexes = []
|
| 347 |
+
for (i, token) in enumerate(tokens):
|
| 348 |
+
if token == "[CLS]" or token == "[SEP]":
|
| 349 |
+
continue
|
| 350 |
+
# Whole Word Masking means that if we mask all of the wordpieces
|
| 351 |
+
# corresponding to an original word. When a word has been split into
|
| 352 |
+
# WordPieces, the first token does not have any marker and any subsequence
|
| 353 |
+
# tokens are prefixed with ##. So whenever we see the ## token, we
|
| 354 |
+
# append it to the previous set of word indexes.
|
| 355 |
+
#
|
| 356 |
+
# Note that Whole Word Masking does *not* change the training code
|
| 357 |
+
# at all -- we still predict each WordPiece independently, softmaxed
|
| 358 |
+
# over the entire vocabulary.
|
| 359 |
+
if (FLAGS.do_whole_word_mask and len(cand_indexes) >= 1 and
|
| 360 |
+
token.startswith("##")):
|
| 361 |
+
cand_indexes[-1].append(i)
|
| 362 |
+
else:
|
| 363 |
+
cand_indexes.append([i])
|
| 364 |
+
|
| 365 |
+
rng.shuffle(cand_indexes)
|
| 366 |
+
|
| 367 |
+
output_tokens = list(tokens)
|
| 368 |
+
|
| 369 |
+
num_to_predict = min(max_predictions_per_seq,
|
| 370 |
+
max(1, int(round(len(tokens) * masked_lm_prob))))
|
| 371 |
+
|
| 372 |
+
masked_lms = []
|
| 373 |
+
covered_indexes = set()
|
| 374 |
+
for index_set in cand_indexes:
|
| 375 |
+
if len(masked_lms) >= num_to_predict:
|
| 376 |
+
break
|
| 377 |
+
# If adding a whole-word mask would exceed the maximum number of
|
| 378 |
+
# predictions, then just skip this candidate.
|
| 379 |
+
if len(masked_lms) + len(index_set) > num_to_predict:
|
| 380 |
+
continue
|
| 381 |
+
is_any_index_covered = False
|
| 382 |
+
for index in index_set:
|
| 383 |
+
if index in covered_indexes:
|
| 384 |
+
is_any_index_covered = True
|
| 385 |
+
break
|
| 386 |
+
if is_any_index_covered:
|
| 387 |
+
continue
|
| 388 |
+
for index in index_set:
|
| 389 |
+
covered_indexes.add(index)
|
| 390 |
+
|
| 391 |
+
masked_token = None
|
| 392 |
+
# 80% of the time, replace with [MASK]
|
| 393 |
+
if rng.random() < 0.8:
|
| 394 |
+
masked_token = "[MASK]"
|
| 395 |
+
else:
|
| 396 |
+
# 10% of the time, keep original
|
| 397 |
+
if rng.random() < 0.5:
|
| 398 |
+
masked_token = tokens[index]
|
| 399 |
+
# 10% of the time, replace with random word
|
| 400 |
+
else:
|
| 401 |
+
masked_token = vocab_words[rng.randint(0, len(vocab_words) - 1)]
|
| 402 |
+
|
| 403 |
+
output_tokens[index] = masked_token
|
| 404 |
+
|
| 405 |
+
masked_lms.append(MaskedLmInstance(index=index, label=tokens[index]))
|
| 406 |
+
assert len(masked_lms) <= num_to_predict
|
| 407 |
+
masked_lms = sorted(masked_lms, key=lambda x: x.index)
|
| 408 |
+
|
| 409 |
+
masked_lm_positions = []
|
| 410 |
+
masked_lm_labels = []
|
| 411 |
+
for p in masked_lms:
|
| 412 |
+
masked_lm_positions.append(p.index)
|
| 413 |
+
masked_lm_labels.append(p.label)
|
| 414 |
+
|
| 415 |
+
return (output_tokens, masked_lm_positions, masked_lm_labels)
|
| 416 |
+
|
| 417 |
+
|
| 418 |
+
def truncate_seq_pair(tokens_a, tokens_b, max_num_tokens, rng):
|
| 419 |
+
"""Truncates a pair of sequences to a maximum sequence length."""
|
| 420 |
+
while True:
|
| 421 |
+
total_length = len(tokens_a) + len(tokens_b)
|
| 422 |
+
if total_length <= max_num_tokens:
|
| 423 |
+
break
|
| 424 |
+
|
| 425 |
+
trunc_tokens = tokens_a if len(tokens_a) > len(tokens_b) else tokens_b
|
| 426 |
+
assert len(trunc_tokens) >= 1
|
| 427 |
+
|
| 428 |
+
# We want to sometimes truncate from the front and sometimes from the
|
| 429 |
+
# back to add more randomness and avoid biases.
|
| 430 |
+
if rng.random() < 0.5:
|
| 431 |
+
del trunc_tokens[0]
|
| 432 |
+
else:
|
| 433 |
+
trunc_tokens.pop()
|
| 434 |
+
|
| 435 |
+
|
| 436 |
+
def main(_):
|
| 437 |
+
tf.logging.set_verbosity(tf.logging.INFO)
|
| 438 |
+
|
| 439 |
+
tokenizer = tokenization.FullTokenizer(
|
| 440 |
+
vocab_file=FLAGS.vocab_file, do_lower_case=FLAGS.do_lower_case)
|
| 441 |
+
|
| 442 |
+
input_files = []
|
| 443 |
+
for input_pattern in FLAGS.input_file.split(","):
|
| 444 |
+
input_files.extend(tf.gfile.Glob(input_pattern))
|
| 445 |
+
|
| 446 |
+
tf.logging.info("*** Reading from input files ***")
|
| 447 |
+
for input_file in input_files:
|
| 448 |
+
tf.logging.info(" %s", input_file)
|
| 449 |
+
|
| 450 |
+
rng = random.Random(FLAGS.random_seed)
|
| 451 |
+
instances = create_training_instances(
|
| 452 |
+
input_files, tokenizer, FLAGS.max_seq_length, FLAGS.dupe_factor,
|
| 453 |
+
FLAGS.short_seq_prob, FLAGS.masked_lm_prob, FLAGS.max_predictions_per_seq,
|
| 454 |
+
rng)
|
| 455 |
+
|
| 456 |
+
output_files = FLAGS.output_file.split(",")
|
| 457 |
+
tf.logging.info("*** Writing to output files ***")
|
| 458 |
+
for output_file in output_files:
|
| 459 |
+
tf.logging.info(" %s", output_file)
|
| 460 |
+
|
| 461 |
+
write_instance_to_example_files(instances, tokenizer, FLAGS.max_seq_length,
|
| 462 |
+
FLAGS.max_predictions_per_seq, output_files)
|
| 463 |
+
|
| 464 |
+
|
| 465 |
+
if __name__ == "__main__":
|
| 466 |
+
flags.mark_flag_as_required("input_file")
|
| 467 |
+
flags.mark_flag_as_required("output_file")
|
| 468 |
+
flags.mark_flag_as_required("vocab_file")
|
| 469 |
+
tf.app.run()
|
bert-master/bert-master/extract_features.py
ADDED
|
@@ -0,0 +1,419 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# coding=utf-8
|
| 2 |
+
# Copyright 2018 The Google AI Language Team Authors.
|
| 3 |
+
#
|
| 4 |
+
# Licensed under the Apache License, Version 2.0 (the "License");
|
| 5 |
+
# you may not use this file except in compliance with the License.
|
| 6 |
+
# You may obtain a copy of the License at
|
| 7 |
+
#
|
| 8 |
+
# http://www.apache.org/licenses/LICENSE-2.0
|
| 9 |
+
#
|
| 10 |
+
# Unless required by applicable law or agreed to in writing, software
|
| 11 |
+
# distributed under the License is distributed on an "AS IS" BASIS,
|
| 12 |
+
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
| 13 |
+
# See the License for the specific language governing permissions and
|
| 14 |
+
# limitations under the License.
|
| 15 |
+
"""Extract pre-computed feature vectors from BERT."""
|
| 16 |
+
|
| 17 |
+
from __future__ import absolute_import
|
| 18 |
+
from __future__ import division
|
| 19 |
+
from __future__ import print_function
|
| 20 |
+
|
| 21 |
+
import codecs
|
| 22 |
+
import collections
|
| 23 |
+
import json
|
| 24 |
+
import re
|
| 25 |
+
|
| 26 |
+
import modeling
|
| 27 |
+
import tokenization
|
| 28 |
+
import tensorflow as tf
|
| 29 |
+
|
| 30 |
+
flags = tf.flags
|
| 31 |
+
|
| 32 |
+
FLAGS = flags.FLAGS
|
| 33 |
+
|
| 34 |
+
flags.DEFINE_string("input_file", None, "")
|
| 35 |
+
|
| 36 |
+
flags.DEFINE_string("output_file", None, "")
|
| 37 |
+
|
| 38 |
+
flags.DEFINE_string("layers", "-1,-2,-3,-4", "")
|
| 39 |
+
|
| 40 |
+
flags.DEFINE_string(
|
| 41 |
+
"bert_config_file", None,
|
| 42 |
+
"The config json file corresponding to the pre-trained BERT model. "
|
| 43 |
+
"This specifies the model architecture.")
|
| 44 |
+
|
| 45 |
+
flags.DEFINE_integer(
|
| 46 |
+
"max_seq_length", 128,
|
| 47 |
+
"The maximum total input sequence length after WordPiece tokenization. "
|
| 48 |
+
"Sequences longer than this will be truncated, and sequences shorter "
|
| 49 |
+
"than this will be padded.")
|
| 50 |
+
|
| 51 |
+
flags.DEFINE_string(
|
| 52 |
+
"init_checkpoint", None,
|
| 53 |
+
"Initial checkpoint (usually from a pre-trained BERT model).")
|
| 54 |
+
|
| 55 |
+
flags.DEFINE_string("vocab_file", None,
|
| 56 |
+
"The vocabulary file that the BERT model was trained on.")
|
| 57 |
+
|
| 58 |
+
flags.DEFINE_bool(
|
| 59 |
+
"do_lower_case", True,
|
| 60 |
+
"Whether to lower case the input text. Should be True for uncased "
|
| 61 |
+
"models and False for cased models.")
|
| 62 |
+
|
| 63 |
+
flags.DEFINE_integer("batch_size", 32, "Batch size for predictions.")
|
| 64 |
+
|
| 65 |
+
flags.DEFINE_bool("use_tpu", False, "Whether to use TPU or GPU/CPU.")
|
| 66 |
+
|
| 67 |
+
flags.DEFINE_string("master", None,
|
| 68 |
+
"If using a TPU, the address of the master.")
|
| 69 |
+
|
| 70 |
+
flags.DEFINE_integer(
|
| 71 |
+
"num_tpu_cores", 8,
|
| 72 |
+
"Only used if `use_tpu` is True. Total number of TPU cores to use.")
|
| 73 |
+
|
| 74 |
+
flags.DEFINE_bool(
|
| 75 |
+
"use_one_hot_embeddings", False,
|
| 76 |
+
"If True, tf.one_hot will be used for embedding lookups, otherwise "
|
| 77 |
+
"tf.nn.embedding_lookup will be used. On TPUs, this should be True "
|
| 78 |
+
"since it is much faster.")
|
| 79 |
+
|
| 80 |
+
|
| 81 |
+
class InputExample(object):
|
| 82 |
+
|
| 83 |
+
def __init__(self, unique_id, text_a, text_b):
|
| 84 |
+
self.unique_id = unique_id
|
| 85 |
+
self.text_a = text_a
|
| 86 |
+
self.text_b = text_b
|
| 87 |
+
|
| 88 |
+
|
| 89 |
+
class InputFeatures(object):
|
| 90 |
+
"""A single set of features of data."""
|
| 91 |
+
|
| 92 |
+
def __init__(self, unique_id, tokens, input_ids, input_mask, input_type_ids):
|
| 93 |
+
self.unique_id = unique_id
|
| 94 |
+
self.tokens = tokens
|
| 95 |
+
self.input_ids = input_ids
|
| 96 |
+
self.input_mask = input_mask
|
| 97 |
+
self.input_type_ids = input_type_ids
|
| 98 |
+
|
| 99 |
+
|
| 100 |
+
def input_fn_builder(features, seq_length):
|
| 101 |
+
"""Creates an `input_fn` closure to be passed to TPUEstimator."""
|
| 102 |
+
|
| 103 |
+
all_unique_ids = []
|
| 104 |
+
all_input_ids = []
|
| 105 |
+
all_input_mask = []
|
| 106 |
+
all_input_type_ids = []
|
| 107 |
+
|
| 108 |
+
for feature in features:
|
| 109 |
+
all_unique_ids.append(feature.unique_id)
|
| 110 |
+
all_input_ids.append(feature.input_ids)
|
| 111 |
+
all_input_mask.append(feature.input_mask)
|
| 112 |
+
all_input_type_ids.append(feature.input_type_ids)
|
| 113 |
+
|
| 114 |
+
def input_fn(params):
|
| 115 |
+
"""The actual input function."""
|
| 116 |
+
batch_size = params["batch_size"]
|
| 117 |
+
|
| 118 |
+
num_examples = len(features)
|
| 119 |
+
|
| 120 |
+
# This is for demo purposes and does NOT scale to large data sets. We do
|
| 121 |
+
# not use Dataset.from_generator() because that uses tf.py_func which is
|
| 122 |
+
# not TPU compatible. The right way to load data is with TFRecordReader.
|
| 123 |
+
d = tf.data.Dataset.from_tensor_slices({
|
| 124 |
+
"unique_ids":
|
| 125 |
+
tf.constant(all_unique_ids, shape=[num_examples], dtype=tf.int32),
|
| 126 |
+
"input_ids":
|
| 127 |
+
tf.constant(
|
| 128 |
+
all_input_ids, shape=[num_examples, seq_length],
|
| 129 |
+
dtype=tf.int32),
|
| 130 |
+
"input_mask":
|
| 131 |
+
tf.constant(
|
| 132 |
+
all_input_mask,
|
| 133 |
+
shape=[num_examples, seq_length],
|
| 134 |
+
dtype=tf.int32),
|
| 135 |
+
"input_type_ids":
|
| 136 |
+
tf.constant(
|
| 137 |
+
all_input_type_ids,
|
| 138 |
+
shape=[num_examples, seq_length],
|
| 139 |
+
dtype=tf.int32),
|
| 140 |
+
})
|
| 141 |
+
|
| 142 |
+
d = d.batch(batch_size=batch_size, drop_remainder=False)
|
| 143 |
+
return d
|
| 144 |
+
|
| 145 |
+
return input_fn
|
| 146 |
+
|
| 147 |
+
|
| 148 |
+
def model_fn_builder(bert_config, init_checkpoint, layer_indexes, use_tpu,
|
| 149 |
+
use_one_hot_embeddings):
|
| 150 |
+
"""Returns `model_fn` closure for TPUEstimator."""
|
| 151 |
+
|
| 152 |
+
def model_fn(features, labels, mode, params): # pylint: disable=unused-argument
|
| 153 |
+
"""The `model_fn` for TPUEstimator."""
|
| 154 |
+
|
| 155 |
+
unique_ids = features["unique_ids"]
|
| 156 |
+
input_ids = features["input_ids"]
|
| 157 |
+
input_mask = features["input_mask"]
|
| 158 |
+
input_type_ids = features["input_type_ids"]
|
| 159 |
+
|
| 160 |
+
model = modeling.BertModel(
|
| 161 |
+
config=bert_config,
|
| 162 |
+
is_training=False,
|
| 163 |
+
input_ids=input_ids,
|
| 164 |
+
input_mask=input_mask,
|
| 165 |
+
token_type_ids=input_type_ids,
|
| 166 |
+
use_one_hot_embeddings=use_one_hot_embeddings)
|
| 167 |
+
|
| 168 |
+
if mode != tf.estimator.ModeKeys.PREDICT:
|
| 169 |
+
raise ValueError("Only PREDICT modes are supported: %s" % (mode))
|
| 170 |
+
|
| 171 |
+
tvars = tf.trainable_variables()
|
| 172 |
+
scaffold_fn = None
|
| 173 |
+
(assignment_map,
|
| 174 |
+
initialized_variable_names) = modeling.get_assignment_map_from_checkpoint(
|
| 175 |
+
tvars, init_checkpoint)
|
| 176 |
+
if use_tpu:
|
| 177 |
+
|
| 178 |
+
def tpu_scaffold():
|
| 179 |
+
tf.train.init_from_checkpoint(init_checkpoint, assignment_map)
|
| 180 |
+
return tf.train.Scaffold()
|
| 181 |
+
|
| 182 |
+
scaffold_fn = tpu_scaffold
|
| 183 |
+
else:
|
| 184 |
+
tf.train.init_from_checkpoint(init_checkpoint, assignment_map)
|
| 185 |
+
|
| 186 |
+
tf.logging.info("**** Trainable Variables ****")
|
| 187 |
+
for var in tvars:
|
| 188 |
+
init_string = ""
|
| 189 |
+
if var.name in initialized_variable_names:
|
| 190 |
+
init_string = ", *INIT_FROM_CKPT*"
|
| 191 |
+
tf.logging.info(" name = %s, shape = %s%s", var.name, var.shape,
|
| 192 |
+
init_string)
|
| 193 |
+
|
| 194 |
+
all_layers = model.get_all_encoder_layers()
|
| 195 |
+
|
| 196 |
+
predictions = {
|
| 197 |
+
"unique_id": unique_ids,
|
| 198 |
+
}
|
| 199 |
+
|
| 200 |
+
for (i, layer_index) in enumerate(layer_indexes):
|
| 201 |
+
predictions["layer_output_%d" % i] = all_layers[layer_index]
|
| 202 |
+
|
| 203 |
+
output_spec = tf.contrib.tpu.TPUEstimatorSpec(
|
| 204 |
+
mode=mode, predictions=predictions, scaffold_fn=scaffold_fn)
|
| 205 |
+
return output_spec
|
| 206 |
+
|
| 207 |
+
return model_fn
|
| 208 |
+
|
| 209 |
+
|
| 210 |
+
def convert_examples_to_features(examples, seq_length, tokenizer):
|
| 211 |
+
"""Loads a data file into a list of `InputBatch`s."""
|
| 212 |
+
|
| 213 |
+
features = []
|
| 214 |
+
for (ex_index, example) in enumerate(examples):
|
| 215 |
+
tokens_a = tokenizer.tokenize(example.text_a)
|
| 216 |
+
|
| 217 |
+
tokens_b = None
|
| 218 |
+
if example.text_b:
|
| 219 |
+
tokens_b = tokenizer.tokenize(example.text_b)
|
| 220 |
+
|
| 221 |
+
if tokens_b:
|
| 222 |
+
# Modifies `tokens_a` and `tokens_b` in place so that the total
|
| 223 |
+
# length is less than the specified length.
|
| 224 |
+
# Account for [CLS], [SEP], [SEP] with "- 3"
|
| 225 |
+
_truncate_seq_pair(tokens_a, tokens_b, seq_length - 3)
|
| 226 |
+
else:
|
| 227 |
+
# Account for [CLS] and [SEP] with "- 2"
|
| 228 |
+
if len(tokens_a) > seq_length - 2:
|
| 229 |
+
tokens_a = tokens_a[0:(seq_length - 2)]
|
| 230 |
+
|
| 231 |
+
# The convention in BERT is:
|
| 232 |
+
# (a) For sequence pairs:
|
| 233 |
+
# tokens: [CLS] is this jack ##son ##ville ? [SEP] no it is not . [SEP]
|
| 234 |
+
# type_ids: 0 0 0 0 0 0 0 0 1 1 1 1 1 1
|
| 235 |
+
# (b) For single sequences:
|
| 236 |
+
# tokens: [CLS] the dog is hairy . [SEP]
|
| 237 |
+
# type_ids: 0 0 0 0 0 0 0
|
| 238 |
+
#
|
| 239 |
+
# Where "type_ids" are used to indicate whether this is the first
|
| 240 |
+
# sequence or the second sequence. The embedding vectors for `type=0` and
|
| 241 |
+
# `type=1` were learned during pre-training and are added to the wordpiece
|
| 242 |
+
# embedding vector (and position vector). This is not *strictly* necessary
|
| 243 |
+
# since the [SEP] token unambiguously separates the sequences, but it makes
|
| 244 |
+
# it easier for the model to learn the concept of sequences.
|
| 245 |
+
#
|
| 246 |
+
# For classification tasks, the first vector (corresponding to [CLS]) is
|
| 247 |
+
# used as as the "sentence vector". Note that this only makes sense because
|
| 248 |
+
# the entire model is fine-tuned.
|
| 249 |
+
tokens = []
|
| 250 |
+
input_type_ids = []
|
| 251 |
+
tokens.append("[CLS]")
|
| 252 |
+
input_type_ids.append(0)
|
| 253 |
+
for token in tokens_a:
|
| 254 |
+
tokens.append(token)
|
| 255 |
+
input_type_ids.append(0)
|
| 256 |
+
tokens.append("[SEP]")
|
| 257 |
+
input_type_ids.append(0)
|
| 258 |
+
|
| 259 |
+
if tokens_b:
|
| 260 |
+
for token in tokens_b:
|
| 261 |
+
tokens.append(token)
|
| 262 |
+
input_type_ids.append(1)
|
| 263 |
+
tokens.append("[SEP]")
|
| 264 |
+
input_type_ids.append(1)
|
| 265 |
+
|
| 266 |
+
input_ids = tokenizer.convert_tokens_to_ids(tokens)
|
| 267 |
+
|
| 268 |
+
# The mask has 1 for real tokens and 0 for padding tokens. Only real
|
| 269 |
+
# tokens are attended to.
|
| 270 |
+
input_mask = [1] * len(input_ids)
|
| 271 |
+
|
| 272 |
+
# Zero-pad up to the sequence length.
|
| 273 |
+
while len(input_ids) < seq_length:
|
| 274 |
+
input_ids.append(0)
|
| 275 |
+
input_mask.append(0)
|
| 276 |
+
input_type_ids.append(0)
|
| 277 |
+
|
| 278 |
+
assert len(input_ids) == seq_length
|
| 279 |
+
assert len(input_mask) == seq_length
|
| 280 |
+
assert len(input_type_ids) == seq_length
|
| 281 |
+
|
| 282 |
+
if ex_index < 5:
|
| 283 |
+
tf.logging.info("*** Example ***")
|
| 284 |
+
tf.logging.info("unique_id: %s" % (example.unique_id))
|
| 285 |
+
tf.logging.info("tokens: %s" % " ".join(
|
| 286 |
+
[tokenization.printable_text(x) for x in tokens]))
|
| 287 |
+
tf.logging.info("input_ids: %s" % " ".join([str(x) for x in input_ids]))
|
| 288 |
+
tf.logging.info("input_mask: %s" % " ".join([str(x) for x in input_mask]))
|
| 289 |
+
tf.logging.info(
|
| 290 |
+
"input_type_ids: %s" % " ".join([str(x) for x in input_type_ids]))
|
| 291 |
+
|
| 292 |
+
features.append(
|
| 293 |
+
InputFeatures(
|
| 294 |
+
unique_id=example.unique_id,
|
| 295 |
+
tokens=tokens,
|
| 296 |
+
input_ids=input_ids,
|
| 297 |
+
input_mask=input_mask,
|
| 298 |
+
input_type_ids=input_type_ids))
|
| 299 |
+
return features
|
| 300 |
+
|
| 301 |
+
|
| 302 |
+
def _truncate_seq_pair(tokens_a, tokens_b, max_length):
|
| 303 |
+
"""Truncates a sequence pair in place to the maximum length."""
|
| 304 |
+
|
| 305 |
+
# This is a simple heuristic which will always truncate the longer sequence
|
| 306 |
+
# one token at a time. This makes more sense than truncating an equal percent
|
| 307 |
+
# of tokens from each, since if one sequence is very short then each token
|
| 308 |
+
# that's truncated likely contains more information than a longer sequence.
|
| 309 |
+
while True:
|
| 310 |
+
total_length = len(tokens_a) + len(tokens_b)
|
| 311 |
+
if total_length <= max_length:
|
| 312 |
+
break
|
| 313 |
+
if len(tokens_a) > len(tokens_b):
|
| 314 |
+
tokens_a.pop()
|
| 315 |
+
else:
|
| 316 |
+
tokens_b.pop()
|
| 317 |
+
|
| 318 |
+
|
| 319 |
+
def read_examples(input_file):
|
| 320 |
+
"""Read a list of `InputExample`s from an input file."""
|
| 321 |
+
examples = []
|
| 322 |
+
unique_id = 0
|
| 323 |
+
with tf.gfile.GFile(input_file, "r") as reader:
|
| 324 |
+
while True:
|
| 325 |
+
line = tokenization.convert_to_unicode(reader.readline())
|
| 326 |
+
if not line:
|
| 327 |
+
break
|
| 328 |
+
line = line.strip()
|
| 329 |
+
text_a = None
|
| 330 |
+
text_b = None
|
| 331 |
+
m = re.match(r"^(.*) \|\|\| (.*)$", line)
|
| 332 |
+
if m is None:
|
| 333 |
+
text_a = line
|
| 334 |
+
else:
|
| 335 |
+
text_a = m.group(1)
|
| 336 |
+
text_b = m.group(2)
|
| 337 |
+
examples.append(
|
| 338 |
+
InputExample(unique_id=unique_id, text_a=text_a, text_b=text_b))
|
| 339 |
+
unique_id += 1
|
| 340 |
+
return examples
|
| 341 |
+
|
| 342 |
+
|
| 343 |
+
def main(_):
|
| 344 |
+
tf.logging.set_verbosity(tf.logging.INFO)
|
| 345 |
+
|
| 346 |
+
layer_indexes = [int(x) for x in FLAGS.layers.split(",")]
|
| 347 |
+
|
| 348 |
+
bert_config = modeling.BertConfig.from_json_file(FLAGS.bert_config_file)
|
| 349 |
+
|
| 350 |
+
tokenizer = tokenization.FullTokenizer(
|
| 351 |
+
vocab_file=FLAGS.vocab_file, do_lower_case=FLAGS.do_lower_case)
|
| 352 |
+
|
| 353 |
+
is_per_host = tf.contrib.tpu.InputPipelineConfig.PER_HOST_V2
|
| 354 |
+
run_config = tf.contrib.tpu.RunConfig(
|
| 355 |
+
master=FLAGS.master,
|
| 356 |
+
tpu_config=tf.contrib.tpu.TPUConfig(
|
| 357 |
+
num_shards=FLAGS.num_tpu_cores,
|
| 358 |
+
per_host_input_for_training=is_per_host))
|
| 359 |
+
|
| 360 |
+
examples = read_examples(FLAGS.input_file)
|
| 361 |
+
|
| 362 |
+
features = convert_examples_to_features(
|
| 363 |
+
examples=examples, seq_length=FLAGS.max_seq_length, tokenizer=tokenizer)
|
| 364 |
+
|
| 365 |
+
unique_id_to_feature = {}
|
| 366 |
+
for feature in features:
|
| 367 |
+
unique_id_to_feature[feature.unique_id] = feature
|
| 368 |
+
|
| 369 |
+
model_fn = model_fn_builder(
|
| 370 |
+
bert_config=bert_config,
|
| 371 |
+
init_checkpoint=FLAGS.init_checkpoint,
|
| 372 |
+
layer_indexes=layer_indexes,
|
| 373 |
+
use_tpu=FLAGS.use_tpu,
|
| 374 |
+
use_one_hot_embeddings=FLAGS.use_one_hot_embeddings)
|
| 375 |
+
|
| 376 |
+
# If TPU is not available, this will fall back to normal Estimator on CPU
|
| 377 |
+
# or GPU.
|
| 378 |
+
estimator = tf.contrib.tpu.TPUEstimator(
|
| 379 |
+
use_tpu=FLAGS.use_tpu,
|
| 380 |
+
model_fn=model_fn,
|
| 381 |
+
config=run_config,
|
| 382 |
+
predict_batch_size=FLAGS.batch_size)
|
| 383 |
+
|
| 384 |
+
input_fn = input_fn_builder(
|
| 385 |
+
features=features, seq_length=FLAGS.max_seq_length)
|
| 386 |
+
|
| 387 |
+
with codecs.getwriter("utf-8")(tf.gfile.Open(FLAGS.output_file,
|
| 388 |
+
"w")) as writer:
|
| 389 |
+
for result in estimator.predict(input_fn, yield_single_examples=True):
|
| 390 |
+
unique_id = int(result["unique_id"])
|
| 391 |
+
feature = unique_id_to_feature[unique_id]
|
| 392 |
+
output_json = collections.OrderedDict()
|
| 393 |
+
output_json["linex_index"] = unique_id
|
| 394 |
+
all_features = []
|
| 395 |
+
for (i, token) in enumerate(feature.tokens):
|
| 396 |
+
all_layers = []
|
| 397 |
+
for (j, layer_index) in enumerate(layer_indexes):
|
| 398 |
+
layer_output = result["layer_output_%d" % j]
|
| 399 |
+
layers = collections.OrderedDict()
|
| 400 |
+
layers["index"] = layer_index
|
| 401 |
+
layers["values"] = [
|
| 402 |
+
round(float(x), 6) for x in layer_output[i:(i + 1)].flat
|
| 403 |
+
]
|
| 404 |
+
all_layers.append(layers)
|
| 405 |
+
features = collections.OrderedDict()
|
| 406 |
+
features["token"] = token
|
| 407 |
+
features["layers"] = all_layers
|
| 408 |
+
all_features.append(features)
|
| 409 |
+
output_json["features"] = all_features
|
| 410 |
+
writer.write(json.dumps(output_json) + "\n")
|
| 411 |
+
|
| 412 |
+
|
| 413 |
+
if __name__ == "__main__":
|
| 414 |
+
flags.mark_flag_as_required("input_file")
|
| 415 |
+
flags.mark_flag_as_required("vocab_file")
|
| 416 |
+
flags.mark_flag_as_required("bert_config_file")
|
| 417 |
+
flags.mark_flag_as_required("init_checkpoint")
|
| 418 |
+
flags.mark_flag_as_required("output_file")
|
| 419 |
+
tf.app.run()
|
bert-master/bert-master/modeling.py
ADDED
|
@@ -0,0 +1,986 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# coding=utf-8
|
| 2 |
+
# Copyright 2018 The Google AI Language Team Authors.
|
| 3 |
+
#
|
| 4 |
+
# Licensed under the Apache License, Version 2.0 (the "License");
|
| 5 |
+
# you may not use this file except in compliance with the License.
|
| 6 |
+
# You may obtain a copy of the License at
|
| 7 |
+
#
|
| 8 |
+
# http://www.apache.org/licenses/LICENSE-2.0
|
| 9 |
+
#
|
| 10 |
+
# Unless required by applicable law or agreed to in writing, software
|
| 11 |
+
# distributed under the License is distributed on an "AS IS" BASIS,
|
| 12 |
+
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
| 13 |
+
# See the License for the specific language governing permissions and
|
| 14 |
+
# limitations under the License.
|
| 15 |
+
"""The main BERT model and related functions."""
|
| 16 |
+
|
| 17 |
+
from __future__ import absolute_import
|
| 18 |
+
from __future__ import division
|
| 19 |
+
from __future__ import print_function
|
| 20 |
+
|
| 21 |
+
import collections
|
| 22 |
+
import copy
|
| 23 |
+
import json
|
| 24 |
+
import math
|
| 25 |
+
import re
|
| 26 |
+
import numpy as np
|
| 27 |
+
import six
|
| 28 |
+
import tensorflow as tf
|
| 29 |
+
|
| 30 |
+
|
| 31 |
+
class BertConfig(object):
|
| 32 |
+
"""Configuration for `BertModel`."""
|
| 33 |
+
|
| 34 |
+
def __init__(self,
|
| 35 |
+
vocab_size,
|
| 36 |
+
hidden_size=768,
|
| 37 |
+
num_hidden_layers=12,
|
| 38 |
+
num_attention_heads=12,
|
| 39 |
+
intermediate_size=3072,
|
| 40 |
+
hidden_act="gelu",
|
| 41 |
+
hidden_dropout_prob=0.1,
|
| 42 |
+
attention_probs_dropout_prob=0.1,
|
| 43 |
+
max_position_embeddings=512,
|
| 44 |
+
type_vocab_size=16,
|
| 45 |
+
initializer_range=0.02):
|
| 46 |
+
"""Constructs BertConfig.
|
| 47 |
+
|
| 48 |
+
Args:
|
| 49 |
+
vocab_size: Vocabulary size of `inputs_ids` in `BertModel`.
|
| 50 |
+
hidden_size: Size of the encoder layers and the pooler layer.
|
| 51 |
+
num_hidden_layers: Number of hidden layers in the Transformer encoder.
|
| 52 |
+
num_attention_heads: Number of attention heads for each attention layer in
|
| 53 |
+
the Transformer encoder.
|
| 54 |
+
intermediate_size: The size of the "intermediate" (i.e., feed-forward)
|
| 55 |
+
layer in the Transformer encoder.
|
| 56 |
+
hidden_act: The non-linear activation function (function or string) in the
|
| 57 |
+
encoder and pooler.
|
| 58 |
+
hidden_dropout_prob: The dropout probability for all fully connected
|
| 59 |
+
layers in the embeddings, encoder, and pooler.
|
| 60 |
+
attention_probs_dropout_prob: The dropout ratio for the attention
|
| 61 |
+
probabilities.
|
| 62 |
+
max_position_embeddings: The maximum sequence length that this model might
|
| 63 |
+
ever be used with. Typically set this to something large just in case
|
| 64 |
+
(e.g., 512 or 1024 or 2048).
|
| 65 |
+
type_vocab_size: The vocabulary size of the `token_type_ids` passed into
|
| 66 |
+
`BertModel`.
|
| 67 |
+
initializer_range: The stdev of the truncated_normal_initializer for
|
| 68 |
+
initializing all weight matrices.
|
| 69 |
+
"""
|
| 70 |
+
self.vocab_size = vocab_size
|
| 71 |
+
self.hidden_size = hidden_size
|
| 72 |
+
self.num_hidden_layers = num_hidden_layers
|
| 73 |
+
self.num_attention_heads = num_attention_heads
|
| 74 |
+
self.hidden_act = hidden_act
|
| 75 |
+
self.intermediate_size = intermediate_size
|
| 76 |
+
self.hidden_dropout_prob = hidden_dropout_prob
|
| 77 |
+
self.attention_probs_dropout_prob = attention_probs_dropout_prob
|
| 78 |
+
self.max_position_embeddings = max_position_embeddings
|
| 79 |
+
self.type_vocab_size = type_vocab_size
|
| 80 |
+
self.initializer_range = initializer_range
|
| 81 |
+
|
| 82 |
+
@classmethod
|
| 83 |
+
def from_dict(cls, json_object):
|
| 84 |
+
"""Constructs a `BertConfig` from a Python dictionary of parameters."""
|
| 85 |
+
config = BertConfig(vocab_size=None)
|
| 86 |
+
for (key, value) in six.iteritems(json_object):
|
| 87 |
+
config.__dict__[key] = value
|
| 88 |
+
return config
|
| 89 |
+
|
| 90 |
+
@classmethod
|
| 91 |
+
def from_json_file(cls, json_file):
|
| 92 |
+
"""Constructs a `BertConfig` from a json file of parameters."""
|
| 93 |
+
with tf.gfile.GFile(json_file, "r") as reader:
|
| 94 |
+
text = reader.read()
|
| 95 |
+
return cls.from_dict(json.loads(text))
|
| 96 |
+
|
| 97 |
+
def to_dict(self):
|
| 98 |
+
"""Serializes this instance to a Python dictionary."""
|
| 99 |
+
output = copy.deepcopy(self.__dict__)
|
| 100 |
+
return output
|
| 101 |
+
|
| 102 |
+
def to_json_string(self):
|
| 103 |
+
"""Serializes this instance to a JSON string."""
|
| 104 |
+
return json.dumps(self.to_dict(), indent=2, sort_keys=True) + "\n"
|
| 105 |
+
|
| 106 |
+
|
| 107 |
+
class BertModel(object):
|
| 108 |
+
"""BERT model ("Bidirectional Encoder Representations from Transformers").
|
| 109 |
+
|
| 110 |
+
Example usage:
|
| 111 |
+
|
| 112 |
+
```python
|
| 113 |
+
# Already been converted into WordPiece token ids
|
| 114 |
+
input_ids = tf.constant([[31, 51, 99], [15, 5, 0]])
|
| 115 |
+
input_mask = tf.constant([[1, 1, 1], [1, 1, 0]])
|
| 116 |
+
token_type_ids = tf.constant([[0, 0, 1], [0, 2, 0]])
|
| 117 |
+
|
| 118 |
+
config = modeling.BertConfig(vocab_size=32000, hidden_size=512,
|
| 119 |
+
num_hidden_layers=8, num_attention_heads=6, intermediate_size=1024)
|
| 120 |
+
|
| 121 |
+
model = modeling.BertModel(config=config, is_training=True,
|
| 122 |
+
input_ids=input_ids, input_mask=input_mask, token_type_ids=token_type_ids)
|
| 123 |
+
|
| 124 |
+
label_embeddings = tf.get_variable(...)
|
| 125 |
+
pooled_output = model.get_pooled_output()
|
| 126 |
+
logits = tf.matmul(pooled_output, label_embeddings)
|
| 127 |
+
...
|
| 128 |
+
```
|
| 129 |
+
"""
|
| 130 |
+
|
| 131 |
+
def __init__(self,
|
| 132 |
+
config,
|
| 133 |
+
is_training,
|
| 134 |
+
input_ids,
|
| 135 |
+
input_mask=None,
|
| 136 |
+
token_type_ids=None,
|
| 137 |
+
use_one_hot_embeddings=False,
|
| 138 |
+
scope=None):
|
| 139 |
+
"""Constructor for BertModel.
|
| 140 |
+
|
| 141 |
+
Args:
|
| 142 |
+
config: `BertConfig` instance.
|
| 143 |
+
is_training: bool. true for training model, false for eval model. Controls
|
| 144 |
+
whether dropout will be applied.
|
| 145 |
+
input_ids: int32 Tensor of shape [batch_size, seq_length].
|
| 146 |
+
input_mask: (optional) int32 Tensor of shape [batch_size, seq_length].
|
| 147 |
+
token_type_ids: (optional) int32 Tensor of shape [batch_size, seq_length].
|
| 148 |
+
use_one_hot_embeddings: (optional) bool. Whether to use one-hot word
|
| 149 |
+
embeddings or tf.embedding_lookup() for the word embeddings.
|
| 150 |
+
scope: (optional) variable scope. Defaults to "bert".
|
| 151 |
+
|
| 152 |
+
Raises:
|
| 153 |
+
ValueError: The config is invalid or one of the input tensor shapes
|
| 154 |
+
is invalid.
|
| 155 |
+
"""
|
| 156 |
+
config = copy.deepcopy(config)
|
| 157 |
+
if not is_training:
|
| 158 |
+
config.hidden_dropout_prob = 0.0
|
| 159 |
+
config.attention_probs_dropout_prob = 0.0
|
| 160 |
+
|
| 161 |
+
input_shape = get_shape_list(input_ids, expected_rank=2)
|
| 162 |
+
batch_size = input_shape[0]
|
| 163 |
+
seq_length = input_shape[1]
|
| 164 |
+
|
| 165 |
+
if input_mask is None:
|
| 166 |
+
input_mask = tf.ones(shape=[batch_size, seq_length], dtype=tf.int32)
|
| 167 |
+
|
| 168 |
+
if token_type_ids is None:
|
| 169 |
+
token_type_ids = tf.zeros(shape=[batch_size, seq_length], dtype=tf.int32)
|
| 170 |
+
|
| 171 |
+
with tf.variable_scope(scope, default_name="bert"):
|
| 172 |
+
with tf.variable_scope("embeddings"):
|
| 173 |
+
# Perform embedding lookup on the word ids.
|
| 174 |
+
(self.embedding_output, self.embedding_table) = embedding_lookup(
|
| 175 |
+
input_ids=input_ids,
|
| 176 |
+
vocab_size=config.vocab_size,
|
| 177 |
+
embedding_size=config.hidden_size,
|
| 178 |
+
initializer_range=config.initializer_range,
|
| 179 |
+
word_embedding_name="word_embeddings",
|
| 180 |
+
use_one_hot_embeddings=use_one_hot_embeddings)
|
| 181 |
+
|
| 182 |
+
# Add positional embeddings and token type embeddings, then layer
|
| 183 |
+
# normalize and perform dropout.
|
| 184 |
+
self.embedding_output = embedding_postprocessor(
|
| 185 |
+
input_tensor=self.embedding_output,
|
| 186 |
+
use_token_type=True,
|
| 187 |
+
token_type_ids=token_type_ids,
|
| 188 |
+
token_type_vocab_size=config.type_vocab_size,
|
| 189 |
+
token_type_embedding_name="token_type_embeddings",
|
| 190 |
+
use_position_embeddings=True,
|
| 191 |
+
position_embedding_name="position_embeddings",
|
| 192 |
+
initializer_range=config.initializer_range,
|
| 193 |
+
max_position_embeddings=config.max_position_embeddings,
|
| 194 |
+
dropout_prob=config.hidden_dropout_prob)
|
| 195 |
+
|
| 196 |
+
with tf.variable_scope("encoder"):
|
| 197 |
+
# This converts a 2D mask of shape [batch_size, seq_length] to a 3D
|
| 198 |
+
# mask of shape [batch_size, seq_length, seq_length] which is used
|
| 199 |
+
# for the attention scores.
|
| 200 |
+
attention_mask = create_attention_mask_from_input_mask(
|
| 201 |
+
input_ids, input_mask)
|
| 202 |
+
|
| 203 |
+
# Run the stacked transformer.
|
| 204 |
+
# `sequence_output` shape = [batch_size, seq_length, hidden_size].
|
| 205 |
+
self.all_encoder_layers = transformer_model(
|
| 206 |
+
input_tensor=self.embedding_output,
|
| 207 |
+
attention_mask=attention_mask,
|
| 208 |
+
hidden_size=config.hidden_size,
|
| 209 |
+
num_hidden_layers=config.num_hidden_layers,
|
| 210 |
+
num_attention_heads=config.num_attention_heads,
|
| 211 |
+
intermediate_size=config.intermediate_size,
|
| 212 |
+
intermediate_act_fn=get_activation(config.hidden_act),
|
| 213 |
+
hidden_dropout_prob=config.hidden_dropout_prob,
|
| 214 |
+
attention_probs_dropout_prob=config.attention_probs_dropout_prob,
|
| 215 |
+
initializer_range=config.initializer_range,
|
| 216 |
+
do_return_all_layers=True)
|
| 217 |
+
|
| 218 |
+
self.sequence_output = self.all_encoder_layers[-1]
|
| 219 |
+
# The "pooler" converts the encoded sequence tensor of shape
|
| 220 |
+
# [batch_size, seq_length, hidden_size] to a tensor of shape
|
| 221 |
+
# [batch_size, hidden_size]. This is necessary for segment-level
|
| 222 |
+
# (or segment-pair-level) classification tasks where we need a fixed
|
| 223 |
+
# dimensional representation of the segment.
|
| 224 |
+
with tf.variable_scope("pooler"):
|
| 225 |
+
# We "pool" the model by simply taking the hidden state corresponding
|
| 226 |
+
# to the first token. We assume that this has been pre-trained
|
| 227 |
+
first_token_tensor = tf.squeeze(self.sequence_output[:, 0:1, :], axis=1)
|
| 228 |
+
self.pooled_output = tf.layers.dense(
|
| 229 |
+
first_token_tensor,
|
| 230 |
+
config.hidden_size,
|
| 231 |
+
activation=tf.tanh,
|
| 232 |
+
kernel_initializer=create_initializer(config.initializer_range))
|
| 233 |
+
|
| 234 |
+
def get_pooled_output(self):
|
| 235 |
+
return self.pooled_output
|
| 236 |
+
|
| 237 |
+
def get_sequence_output(self):
|
| 238 |
+
"""Gets final hidden layer of encoder.
|
| 239 |
+
|
| 240 |
+
Returns:
|
| 241 |
+
float Tensor of shape [batch_size, seq_length, hidden_size] corresponding
|
| 242 |
+
to the final hidden of the transformer encoder.
|
| 243 |
+
"""
|
| 244 |
+
return self.sequence_output
|
| 245 |
+
|
| 246 |
+
def get_all_encoder_layers(self):
|
| 247 |
+
return self.all_encoder_layers
|
| 248 |
+
|
| 249 |
+
def get_embedding_output(self):
|
| 250 |
+
"""Gets output of the embedding lookup (i.e., input to the transformer).
|
| 251 |
+
|
| 252 |
+
Returns:
|
| 253 |
+
float Tensor of shape [batch_size, seq_length, hidden_size] corresponding
|
| 254 |
+
to the output of the embedding layer, after summing the word
|
| 255 |
+
embeddings with the positional embeddings and the token type embeddings,
|
| 256 |
+
then performing layer normalization. This is the input to the transformer.
|
| 257 |
+
"""
|
| 258 |
+
return self.embedding_output
|
| 259 |
+
|
| 260 |
+
def get_embedding_table(self):
|
| 261 |
+
return self.embedding_table
|
| 262 |
+
|
| 263 |
+
|
| 264 |
+
def gelu(x):
|
| 265 |
+
"""Gaussian Error Linear Unit.
|
| 266 |
+
|
| 267 |
+
This is a smoother version of the RELU.
|
| 268 |
+
Original paper: https://arxiv.org/abs/1606.08415
|
| 269 |
+
Args:
|
| 270 |
+
x: float Tensor to perform activation.
|
| 271 |
+
|
| 272 |
+
Returns:
|
| 273 |
+
`x` with the GELU activation applied.
|
| 274 |
+
"""
|
| 275 |
+
cdf = 0.5 * (1.0 + tf.tanh(
|
| 276 |
+
(np.sqrt(2 / np.pi) * (x + 0.044715 * tf.pow(x, 3)))))
|
| 277 |
+
return x * cdf
|
| 278 |
+
|
| 279 |
+
|
| 280 |
+
def get_activation(activation_string):
|
| 281 |
+
"""Maps a string to a Python function, e.g., "relu" => `tf.nn.relu`.
|
| 282 |
+
|
| 283 |
+
Args:
|
| 284 |
+
activation_string: String name of the activation function.
|
| 285 |
+
|
| 286 |
+
Returns:
|
| 287 |
+
A Python function corresponding to the activation function. If
|
| 288 |
+
`activation_string` is None, empty, or "linear", this will return None.
|
| 289 |
+
If `activation_string` is not a string, it will return `activation_string`.
|
| 290 |
+
|
| 291 |
+
Raises:
|
| 292 |
+
ValueError: The `activation_string` does not correspond to a known
|
| 293 |
+
activation.
|
| 294 |
+
"""
|
| 295 |
+
|
| 296 |
+
# We assume that anything that"s not a string is already an activation
|
| 297 |
+
# function, so we just return it.
|
| 298 |
+
if not isinstance(activation_string, six.string_types):
|
| 299 |
+
return activation_string
|
| 300 |
+
|
| 301 |
+
if not activation_string:
|
| 302 |
+
return None
|
| 303 |
+
|
| 304 |
+
act = activation_string.lower()
|
| 305 |
+
if act == "linear":
|
| 306 |
+
return None
|
| 307 |
+
elif act == "relu":
|
| 308 |
+
return tf.nn.relu
|
| 309 |
+
elif act == "gelu":
|
| 310 |
+
return gelu
|
| 311 |
+
elif act == "tanh":
|
| 312 |
+
return tf.tanh
|
| 313 |
+
else:
|
| 314 |
+
raise ValueError("Unsupported activation: %s" % act)
|
| 315 |
+
|
| 316 |
+
|
| 317 |
+
def get_assignment_map_from_checkpoint(tvars, init_checkpoint):
|
| 318 |
+
"""Compute the union of the current variables and checkpoint variables."""
|
| 319 |
+
assignment_map = {}
|
| 320 |
+
initialized_variable_names = {}
|
| 321 |
+
|
| 322 |
+
name_to_variable = collections.OrderedDict()
|
| 323 |
+
for var in tvars:
|
| 324 |
+
name = var.name
|
| 325 |
+
m = re.match("^(.*):\\d+$", name)
|
| 326 |
+
if m is not None:
|
| 327 |
+
name = m.group(1)
|
| 328 |
+
name_to_variable[name] = var
|
| 329 |
+
|
| 330 |
+
init_vars = tf.train.list_variables(init_checkpoint)
|
| 331 |
+
|
| 332 |
+
assignment_map = collections.OrderedDict()
|
| 333 |
+
for x in init_vars:
|
| 334 |
+
(name, var) = (x[0], x[1])
|
| 335 |
+
if name not in name_to_variable:
|
| 336 |
+
continue
|
| 337 |
+
assignment_map[name] = name
|
| 338 |
+
initialized_variable_names[name] = 1
|
| 339 |
+
initialized_variable_names[name + ":0"] = 1
|
| 340 |
+
|
| 341 |
+
return (assignment_map, initialized_variable_names)
|
| 342 |
+
|
| 343 |
+
|
| 344 |
+
def dropout(input_tensor, dropout_prob):
|
| 345 |
+
"""Perform dropout.
|
| 346 |
+
|
| 347 |
+
Args:
|
| 348 |
+
input_tensor: float Tensor.
|
| 349 |
+
dropout_prob: Python float. The probability of dropping out a value (NOT of
|
| 350 |
+
*keeping* a dimension as in `tf.nn.dropout`).
|
| 351 |
+
|
| 352 |
+
Returns:
|
| 353 |
+
A version of `input_tensor` with dropout applied.
|
| 354 |
+
"""
|
| 355 |
+
if dropout_prob is None or dropout_prob == 0.0:
|
| 356 |
+
return input_tensor
|
| 357 |
+
|
| 358 |
+
output = tf.nn.dropout(input_tensor, 1.0 - dropout_prob)
|
| 359 |
+
return output
|
| 360 |
+
|
| 361 |
+
|
| 362 |
+
def layer_norm(input_tensor, name=None):
|
| 363 |
+
"""Run layer normalization on the last dimension of the tensor."""
|
| 364 |
+
return tf.contrib.layers.layer_norm(
|
| 365 |
+
inputs=input_tensor, begin_norm_axis=-1, begin_params_axis=-1, scope=name)
|
| 366 |
+
|
| 367 |
+
|
| 368 |
+
def layer_norm_and_dropout(input_tensor, dropout_prob, name=None):
|
| 369 |
+
"""Runs layer normalization followed by dropout."""
|
| 370 |
+
output_tensor = layer_norm(input_tensor, name)
|
| 371 |
+
output_tensor = dropout(output_tensor, dropout_prob)
|
| 372 |
+
return output_tensor
|
| 373 |
+
|
| 374 |
+
|
| 375 |
+
def create_initializer(initializer_range=0.02):
|
| 376 |
+
"""Creates a `truncated_normal_initializer` with the given range."""
|
| 377 |
+
return tf.truncated_normal_initializer(stddev=initializer_range)
|
| 378 |
+
|
| 379 |
+
|
| 380 |
+
def embedding_lookup(input_ids,
|
| 381 |
+
vocab_size,
|
| 382 |
+
embedding_size=128,
|
| 383 |
+
initializer_range=0.02,
|
| 384 |
+
word_embedding_name="word_embeddings",
|
| 385 |
+
use_one_hot_embeddings=False):
|
| 386 |
+
"""Looks up words embeddings for id tensor.
|
| 387 |
+
|
| 388 |
+
Args:
|
| 389 |
+
input_ids: int32 Tensor of shape [batch_size, seq_length] containing word
|
| 390 |
+
ids.
|
| 391 |
+
vocab_size: int. Size of the embedding vocabulary.
|
| 392 |
+
embedding_size: int. Width of the word embeddings.
|
| 393 |
+
initializer_range: float. Embedding initialization range.
|
| 394 |
+
word_embedding_name: string. Name of the embedding table.
|
| 395 |
+
use_one_hot_embeddings: bool. If True, use one-hot method for word
|
| 396 |
+
embeddings. If False, use `tf.gather()`.
|
| 397 |
+
|
| 398 |
+
Returns:
|
| 399 |
+
float Tensor of shape [batch_size, seq_length, embedding_size].
|
| 400 |
+
"""
|
| 401 |
+
# This function assumes that the input is of shape [batch_size, seq_length,
|
| 402 |
+
# num_inputs].
|
| 403 |
+
#
|
| 404 |
+
# If the input is a 2D tensor of shape [batch_size, seq_length], we
|
| 405 |
+
# reshape to [batch_size, seq_length, 1].
|
| 406 |
+
if input_ids.shape.ndims == 2:
|
| 407 |
+
input_ids = tf.expand_dims(input_ids, axis=[-1])
|
| 408 |
+
|
| 409 |
+
embedding_table = tf.get_variable(
|
| 410 |
+
name=word_embedding_name,
|
| 411 |
+
shape=[vocab_size, embedding_size],
|
| 412 |
+
initializer=create_initializer(initializer_range))
|
| 413 |
+
|
| 414 |
+
flat_input_ids = tf.reshape(input_ids, [-1])
|
| 415 |
+
if use_one_hot_embeddings:
|
| 416 |
+
one_hot_input_ids = tf.one_hot(flat_input_ids, depth=vocab_size)
|
| 417 |
+
output = tf.matmul(one_hot_input_ids, embedding_table)
|
| 418 |
+
else:
|
| 419 |
+
output = tf.gather(embedding_table, flat_input_ids)
|
| 420 |
+
|
| 421 |
+
input_shape = get_shape_list(input_ids)
|
| 422 |
+
|
| 423 |
+
output = tf.reshape(output,
|
| 424 |
+
input_shape[0:-1] + [input_shape[-1] * embedding_size])
|
| 425 |
+
return (output, embedding_table)
|
| 426 |
+
|
| 427 |
+
|
| 428 |
+
def embedding_postprocessor(input_tensor,
|
| 429 |
+
use_token_type=False,
|
| 430 |
+
token_type_ids=None,
|
| 431 |
+
token_type_vocab_size=16,
|
| 432 |
+
token_type_embedding_name="token_type_embeddings",
|
| 433 |
+
use_position_embeddings=True,
|
| 434 |
+
position_embedding_name="position_embeddings",
|
| 435 |
+
initializer_range=0.02,
|
| 436 |
+
max_position_embeddings=512,
|
| 437 |
+
dropout_prob=0.1):
|
| 438 |
+
"""Performs various post-processing on a word embedding tensor.
|
| 439 |
+
|
| 440 |
+
Args:
|
| 441 |
+
input_tensor: float Tensor of shape [batch_size, seq_length,
|
| 442 |
+
embedding_size].
|
| 443 |
+
use_token_type: bool. Whether to add embeddings for `token_type_ids`.
|
| 444 |
+
token_type_ids: (optional) int32 Tensor of shape [batch_size, seq_length].
|
| 445 |
+
Must be specified if `use_token_type` is True.
|
| 446 |
+
token_type_vocab_size: int. The vocabulary size of `token_type_ids`.
|
| 447 |
+
token_type_embedding_name: string. The name of the embedding table variable
|
| 448 |
+
for token type ids.
|
| 449 |
+
use_position_embeddings: bool. Whether to add position embeddings for the
|
| 450 |
+
position of each token in the sequence.
|
| 451 |
+
position_embedding_name: string. The name of the embedding table variable
|
| 452 |
+
for positional embeddings.
|
| 453 |
+
initializer_range: float. Range of the weight initialization.
|
| 454 |
+
max_position_embeddings: int. Maximum sequence length that might ever be
|
| 455 |
+
used with this model. This can be longer than the sequence length of
|
| 456 |
+
input_tensor, but cannot be shorter.
|
| 457 |
+
dropout_prob: float. Dropout probability applied to the final output tensor.
|
| 458 |
+
|
| 459 |
+
Returns:
|
| 460 |
+
float tensor with same shape as `input_tensor`.
|
| 461 |
+
|
| 462 |
+
Raises:
|
| 463 |
+
ValueError: One of the tensor shapes or input values is invalid.
|
| 464 |
+
"""
|
| 465 |
+
input_shape = get_shape_list(input_tensor, expected_rank=3)
|
| 466 |
+
batch_size = input_shape[0]
|
| 467 |
+
seq_length = input_shape[1]
|
| 468 |
+
width = input_shape[2]
|
| 469 |
+
|
| 470 |
+
output = input_tensor
|
| 471 |
+
|
| 472 |
+
if use_token_type:
|
| 473 |
+
if token_type_ids is None:
|
| 474 |
+
raise ValueError("`token_type_ids` must be specified if"
|
| 475 |
+
"`use_token_type` is True.")
|
| 476 |
+
token_type_table = tf.get_variable(
|
| 477 |
+
name=token_type_embedding_name,
|
| 478 |
+
shape=[token_type_vocab_size, width],
|
| 479 |
+
initializer=create_initializer(initializer_range))
|
| 480 |
+
# This vocab will be small so we always do one-hot here, since it is always
|
| 481 |
+
# faster for a small vocabulary.
|
| 482 |
+
flat_token_type_ids = tf.reshape(token_type_ids, [-1])
|
| 483 |
+
one_hot_ids = tf.one_hot(flat_token_type_ids, depth=token_type_vocab_size)
|
| 484 |
+
token_type_embeddings = tf.matmul(one_hot_ids, token_type_table)
|
| 485 |
+
token_type_embeddings = tf.reshape(token_type_embeddings,
|
| 486 |
+
[batch_size, seq_length, width])
|
| 487 |
+
output += token_type_embeddings
|
| 488 |
+
|
| 489 |
+
if use_position_embeddings:
|
| 490 |
+
assert_op = tf.assert_less_equal(seq_length, max_position_embeddings)
|
| 491 |
+
with tf.control_dependencies([assert_op]):
|
| 492 |
+
full_position_embeddings = tf.get_variable(
|
| 493 |
+
name=position_embedding_name,
|
| 494 |
+
shape=[max_position_embeddings, width],
|
| 495 |
+
initializer=create_initializer(initializer_range))
|
| 496 |
+
# Since the position embedding table is a learned variable, we create it
|
| 497 |
+
# using a (long) sequence length `max_position_embeddings`. The actual
|
| 498 |
+
# sequence length might be shorter than this, for faster training of
|
| 499 |
+
# tasks that do not have long sequences.
|
| 500 |
+
#
|
| 501 |
+
# So `full_position_embeddings` is effectively an embedding table
|
| 502 |
+
# for position [0, 1, 2, ..., max_position_embeddings-1], and the current
|
| 503 |
+
# sequence has positions [0, 1, 2, ... seq_length-1], so we can just
|
| 504 |
+
# perform a slice.
|
| 505 |
+
position_embeddings = tf.slice(full_position_embeddings, [0, 0],
|
| 506 |
+
[seq_length, -1])
|
| 507 |
+
num_dims = len(output.shape.as_list())
|
| 508 |
+
|
| 509 |
+
# Only the last two dimensions are relevant (`seq_length` and `width`), so
|
| 510 |
+
# we broadcast among the first dimensions, which is typically just
|
| 511 |
+
# the batch size.
|
| 512 |
+
position_broadcast_shape = []
|
| 513 |
+
for _ in range(num_dims - 2):
|
| 514 |
+
position_broadcast_shape.append(1)
|
| 515 |
+
position_broadcast_shape.extend([seq_length, width])
|
| 516 |
+
position_embeddings = tf.reshape(position_embeddings,
|
| 517 |
+
position_broadcast_shape)
|
| 518 |
+
output += position_embeddings
|
| 519 |
+
|
| 520 |
+
output = layer_norm_and_dropout(output, dropout_prob)
|
| 521 |
+
return output
|
| 522 |
+
|
| 523 |
+
|
| 524 |
+
def create_attention_mask_from_input_mask(from_tensor, to_mask):
|
| 525 |
+
"""Create 3D attention mask from a 2D tensor mask.
|
| 526 |
+
|
| 527 |
+
Args:
|
| 528 |
+
from_tensor: 2D or 3D Tensor of shape [batch_size, from_seq_length, ...].
|
| 529 |
+
to_mask: int32 Tensor of shape [batch_size, to_seq_length].
|
| 530 |
+
|
| 531 |
+
Returns:
|
| 532 |
+
float Tensor of shape [batch_size, from_seq_length, to_seq_length].
|
| 533 |
+
"""
|
| 534 |
+
from_shape = get_shape_list(from_tensor, expected_rank=[2, 3])
|
| 535 |
+
batch_size = from_shape[0]
|
| 536 |
+
from_seq_length = from_shape[1]
|
| 537 |
+
|
| 538 |
+
to_shape = get_shape_list(to_mask, expected_rank=2)
|
| 539 |
+
to_seq_length = to_shape[1]
|
| 540 |
+
|
| 541 |
+
to_mask = tf.cast(
|
| 542 |
+
tf.reshape(to_mask, [batch_size, 1, to_seq_length]), tf.float32)
|
| 543 |
+
|
| 544 |
+
# We don't assume that `from_tensor` is a mask (although it could be). We
|
| 545 |
+
# don't actually care if we attend *from* padding tokens (only *to* padding)
|
| 546 |
+
# tokens so we create a tensor of all ones.
|
| 547 |
+
#
|
| 548 |
+
# `broadcast_ones` = [batch_size, from_seq_length, 1]
|
| 549 |
+
broadcast_ones = tf.ones(
|
| 550 |
+
shape=[batch_size, from_seq_length, 1], dtype=tf.float32)
|
| 551 |
+
|
| 552 |
+
# Here we broadcast along two dimensions to create the mask.
|
| 553 |
+
mask = broadcast_ones * to_mask
|
| 554 |
+
|
| 555 |
+
return mask
|
| 556 |
+
|
| 557 |
+
|
| 558 |
+
def attention_layer(from_tensor,
|
| 559 |
+
to_tensor,
|
| 560 |
+
attention_mask=None,
|
| 561 |
+
num_attention_heads=1,
|
| 562 |
+
size_per_head=512,
|
| 563 |
+
query_act=None,
|
| 564 |
+
key_act=None,
|
| 565 |
+
value_act=None,
|
| 566 |
+
attention_probs_dropout_prob=0.0,
|
| 567 |
+
initializer_range=0.02,
|
| 568 |
+
do_return_2d_tensor=False,
|
| 569 |
+
batch_size=None,
|
| 570 |
+
from_seq_length=None,
|
| 571 |
+
to_seq_length=None):
|
| 572 |
+
"""Performs multi-headed attention from `from_tensor` to `to_tensor`.
|
| 573 |
+
|
| 574 |
+
This is an implementation of multi-headed attention based on "Attention
|
| 575 |
+
is all you Need". If `from_tensor` and `to_tensor` are the same, then
|
| 576 |
+
this is self-attention. Each timestep in `from_tensor` attends to the
|
| 577 |
+
corresponding sequence in `to_tensor`, and returns a fixed-with vector.
|
| 578 |
+
|
| 579 |
+
This function first projects `from_tensor` into a "query" tensor and
|
| 580 |
+
`to_tensor` into "key" and "value" tensors. These are (effectively) a list
|
| 581 |
+
of tensors of length `num_attention_heads`, where each tensor is of shape
|
| 582 |
+
[batch_size, seq_length, size_per_head].
|
| 583 |
+
|
| 584 |
+
Then, the query and key tensors are dot-producted and scaled. These are
|
| 585 |
+
softmaxed to obtain attention probabilities. The value tensors are then
|
| 586 |
+
interpolated by these probabilities, then concatenated back to a single
|
| 587 |
+
tensor and returned.
|
| 588 |
+
|
| 589 |
+
In practice, the multi-headed attention are done with transposes and
|
| 590 |
+
reshapes rather than actual separate tensors.
|
| 591 |
+
|
| 592 |
+
Args:
|
| 593 |
+
from_tensor: float Tensor of shape [batch_size, from_seq_length,
|
| 594 |
+
from_width].
|
| 595 |
+
to_tensor: float Tensor of shape [batch_size, to_seq_length, to_width].
|
| 596 |
+
attention_mask: (optional) int32 Tensor of shape [batch_size,
|
| 597 |
+
from_seq_length, to_seq_length]. The values should be 1 or 0. The
|
| 598 |
+
attention scores will effectively be set to -infinity for any positions in
|
| 599 |
+
the mask that are 0, and will be unchanged for positions that are 1.
|
| 600 |
+
num_attention_heads: int. Number of attention heads.
|
| 601 |
+
size_per_head: int. Size of each attention head.
|
| 602 |
+
query_act: (optional) Activation function for the query transform.
|
| 603 |
+
key_act: (optional) Activation function for the key transform.
|
| 604 |
+
value_act: (optional) Activation function for the value transform.
|
| 605 |
+
attention_probs_dropout_prob: (optional) float. Dropout probability of the
|
| 606 |
+
attention probabilities.
|
| 607 |
+
initializer_range: float. Range of the weight initializer.
|
| 608 |
+
do_return_2d_tensor: bool. If True, the output will be of shape [batch_size
|
| 609 |
+
* from_seq_length, num_attention_heads * size_per_head]. If False, the
|
| 610 |
+
output will be of shape [batch_size, from_seq_length, num_attention_heads
|
| 611 |
+
* size_per_head].
|
| 612 |
+
batch_size: (Optional) int. If the input is 2D, this might be the batch size
|
| 613 |
+
of the 3D version of the `from_tensor` and `to_tensor`.
|
| 614 |
+
from_seq_length: (Optional) If the input is 2D, this might be the seq length
|
| 615 |
+
of the 3D version of the `from_tensor`.
|
| 616 |
+
to_seq_length: (Optional) If the input is 2D, this might be the seq length
|
| 617 |
+
of the 3D version of the `to_tensor`.
|
| 618 |
+
|
| 619 |
+
Returns:
|
| 620 |
+
float Tensor of shape [batch_size, from_seq_length,
|
| 621 |
+
num_attention_heads * size_per_head]. (If `do_return_2d_tensor` is
|
| 622 |
+
true, this will be of shape [batch_size * from_seq_length,
|
| 623 |
+
num_attention_heads * size_per_head]).
|
| 624 |
+
|
| 625 |
+
Raises:
|
| 626 |
+
ValueError: Any of the arguments or tensor shapes are invalid.
|
| 627 |
+
"""
|
| 628 |
+
|
| 629 |
+
def transpose_for_scores(input_tensor, batch_size, num_attention_heads,
|
| 630 |
+
seq_length, width):
|
| 631 |
+
output_tensor = tf.reshape(
|
| 632 |
+
input_tensor, [batch_size, seq_length, num_attention_heads, width])
|
| 633 |
+
|
| 634 |
+
output_tensor = tf.transpose(output_tensor, [0, 2, 1, 3])
|
| 635 |
+
return output_tensor
|
| 636 |
+
|
| 637 |
+
from_shape = get_shape_list(from_tensor, expected_rank=[2, 3])
|
| 638 |
+
to_shape = get_shape_list(to_tensor, expected_rank=[2, 3])
|
| 639 |
+
|
| 640 |
+
if len(from_shape) != len(to_shape):
|
| 641 |
+
raise ValueError(
|
| 642 |
+
"The rank of `from_tensor` must match the rank of `to_tensor`.")
|
| 643 |
+
|
| 644 |
+
if len(from_shape) == 3:
|
| 645 |
+
batch_size = from_shape[0]
|
| 646 |
+
from_seq_length = from_shape[1]
|
| 647 |
+
to_seq_length = to_shape[1]
|
| 648 |
+
elif len(from_shape) == 2:
|
| 649 |
+
if (batch_size is None or from_seq_length is None or to_seq_length is None):
|
| 650 |
+
raise ValueError(
|
| 651 |
+
"When passing in rank 2 tensors to attention_layer, the values "
|
| 652 |
+
"for `batch_size`, `from_seq_length`, and `to_seq_length` "
|
| 653 |
+
"must all be specified.")
|
| 654 |
+
|
| 655 |
+
# Scalar dimensions referenced here:
|
| 656 |
+
# B = batch size (number of sequences)
|
| 657 |
+
# F = `from_tensor` sequence length
|
| 658 |
+
# T = `to_tensor` sequence length
|
| 659 |
+
# N = `num_attention_heads`
|
| 660 |
+
# H = `size_per_head`
|
| 661 |
+
|
| 662 |
+
from_tensor_2d = reshape_to_matrix(from_tensor)
|
| 663 |
+
to_tensor_2d = reshape_to_matrix(to_tensor)
|
| 664 |
+
|
| 665 |
+
# `query_layer` = [B*F, N*H]
|
| 666 |
+
query_layer = tf.layers.dense(
|
| 667 |
+
from_tensor_2d,
|
| 668 |
+
num_attention_heads * size_per_head,
|
| 669 |
+
activation=query_act,
|
| 670 |
+
name="query",
|
| 671 |
+
kernel_initializer=create_initializer(initializer_range))
|
| 672 |
+
|
| 673 |
+
# `key_layer` = [B*T, N*H]
|
| 674 |
+
key_layer = tf.layers.dense(
|
| 675 |
+
to_tensor_2d,
|
| 676 |
+
num_attention_heads * size_per_head,
|
| 677 |
+
activation=key_act,
|
| 678 |
+
name="key",
|
| 679 |
+
kernel_initializer=create_initializer(initializer_range))
|
| 680 |
+
|
| 681 |
+
# `value_layer` = [B*T, N*H]
|
| 682 |
+
value_layer = tf.layers.dense(
|
| 683 |
+
to_tensor_2d,
|
| 684 |
+
num_attention_heads * size_per_head,
|
| 685 |
+
activation=value_act,
|
| 686 |
+
name="value",
|
| 687 |
+
kernel_initializer=create_initializer(initializer_range))
|
| 688 |
+
|
| 689 |
+
# `query_layer` = [B, N, F, H]
|
| 690 |
+
query_layer = transpose_for_scores(query_layer, batch_size,
|
| 691 |
+
num_attention_heads, from_seq_length,
|
| 692 |
+
size_per_head)
|
| 693 |
+
|
| 694 |
+
# `key_layer` = [B, N, T, H]
|
| 695 |
+
key_layer = transpose_for_scores(key_layer, batch_size, num_attention_heads,
|
| 696 |
+
to_seq_length, size_per_head)
|
| 697 |
+
|
| 698 |
+
# Take the dot product between "query" and "key" to get the raw
|
| 699 |
+
# attention scores.
|
| 700 |
+
# `attention_scores` = [B, N, F, T]
|
| 701 |
+
attention_scores = tf.matmul(query_layer, key_layer, transpose_b=True)
|
| 702 |
+
attention_scores = tf.multiply(attention_scores,
|
| 703 |
+
1.0 / math.sqrt(float(size_per_head)))
|
| 704 |
+
|
| 705 |
+
if attention_mask is not None:
|
| 706 |
+
# `attention_mask` = [B, 1, F, T]
|
| 707 |
+
attention_mask = tf.expand_dims(attention_mask, axis=[1])
|
| 708 |
+
|
| 709 |
+
# Since attention_mask is 1.0 for positions we want to attend and 0.0 for
|
| 710 |
+
# masked positions, this operation will create a tensor which is 0.0 for
|
| 711 |
+
# positions we want to attend and -10000.0 for masked positions.
|
| 712 |
+
adder = (1.0 - tf.cast(attention_mask, tf.float32)) * -10000.0
|
| 713 |
+
|
| 714 |
+
# Since we are adding it to the raw scores before the softmax, this is
|
| 715 |
+
# effectively the same as removing these entirely.
|
| 716 |
+
attention_scores += adder
|
| 717 |
+
|
| 718 |
+
# Normalize the attention scores to probabilities.
|
| 719 |
+
# `attention_probs` = [B, N, F, T]
|
| 720 |
+
attention_probs = tf.nn.softmax(attention_scores)
|
| 721 |
+
|
| 722 |
+
# This is actually dropping out entire tokens to attend to, which might
|
| 723 |
+
# seem a bit unusual, but is taken from the original Transformer paper.
|
| 724 |
+
attention_probs = dropout(attention_probs, attention_probs_dropout_prob)
|
| 725 |
+
|
| 726 |
+
# `value_layer` = [B, T, N, H]
|
| 727 |
+
value_layer = tf.reshape(
|
| 728 |
+
value_layer,
|
| 729 |
+
[batch_size, to_seq_length, num_attention_heads, size_per_head])
|
| 730 |
+
|
| 731 |
+
# `value_layer` = [B, N, T, H]
|
| 732 |
+
value_layer = tf.transpose(value_layer, [0, 2, 1, 3])
|
| 733 |
+
|
| 734 |
+
# `context_layer` = [B, N, F, H]
|
| 735 |
+
context_layer = tf.matmul(attention_probs, value_layer)
|
| 736 |
+
|
| 737 |
+
# `context_layer` = [B, F, N, H]
|
| 738 |
+
context_layer = tf.transpose(context_layer, [0, 2, 1, 3])
|
| 739 |
+
|
| 740 |
+
if do_return_2d_tensor:
|
| 741 |
+
# `context_layer` = [B*F, N*H]
|
| 742 |
+
context_layer = tf.reshape(
|
| 743 |
+
context_layer,
|
| 744 |
+
[batch_size * from_seq_length, num_attention_heads * size_per_head])
|
| 745 |
+
else:
|
| 746 |
+
# `context_layer` = [B, F, N*H]
|
| 747 |
+
context_layer = tf.reshape(
|
| 748 |
+
context_layer,
|
| 749 |
+
[batch_size, from_seq_length, num_attention_heads * size_per_head])
|
| 750 |
+
|
| 751 |
+
return context_layer
|
| 752 |
+
|
| 753 |
+
|
| 754 |
+
def transformer_model(input_tensor,
|
| 755 |
+
attention_mask=None,
|
| 756 |
+
hidden_size=768,
|
| 757 |
+
num_hidden_layers=12,
|
| 758 |
+
num_attention_heads=12,
|
| 759 |
+
intermediate_size=3072,
|
| 760 |
+
intermediate_act_fn=gelu,
|
| 761 |
+
hidden_dropout_prob=0.1,
|
| 762 |
+
attention_probs_dropout_prob=0.1,
|
| 763 |
+
initializer_range=0.02,
|
| 764 |
+
do_return_all_layers=False):
|
| 765 |
+
"""Multi-headed, multi-layer Transformer from "Attention is All You Need".
|
| 766 |
+
|
| 767 |
+
This is almost an exact implementation of the original Transformer encoder.
|
| 768 |
+
|
| 769 |
+
See the original paper:
|
| 770 |
+
https://arxiv.org/abs/1706.03762
|
| 771 |
+
|
| 772 |
+
Also see:
|
| 773 |
+
https://github.com/tensorflow/tensor2tensor/blob/master/tensor2tensor/models/transformer.py
|
| 774 |
+
|
| 775 |
+
Args:
|
| 776 |
+
input_tensor: float Tensor of shape [batch_size, seq_length, hidden_size].
|
| 777 |
+
attention_mask: (optional) int32 Tensor of shape [batch_size, seq_length,
|
| 778 |
+
seq_length], with 1 for positions that can be attended to and 0 in
|
| 779 |
+
positions that should not be.
|
| 780 |
+
hidden_size: int. Hidden size of the Transformer.
|
| 781 |
+
num_hidden_layers: int. Number of layers (blocks) in the Transformer.
|
| 782 |
+
num_attention_heads: int. Number of attention heads in the Transformer.
|
| 783 |
+
intermediate_size: int. The size of the "intermediate" (a.k.a., feed
|
| 784 |
+
forward) layer.
|
| 785 |
+
intermediate_act_fn: function. The non-linear activation function to apply
|
| 786 |
+
to the output of the intermediate/feed-forward layer.
|
| 787 |
+
hidden_dropout_prob: float. Dropout probability for the hidden layers.
|
| 788 |
+
attention_probs_dropout_prob: float. Dropout probability of the attention
|
| 789 |
+
probabilities.
|
| 790 |
+
initializer_range: float. Range of the initializer (stddev of truncated
|
| 791 |
+
normal).
|
| 792 |
+
do_return_all_layers: Whether to also return all layers or just the final
|
| 793 |
+
layer.
|
| 794 |
+
|
| 795 |
+
Returns:
|
| 796 |
+
float Tensor of shape [batch_size, seq_length, hidden_size], the final
|
| 797 |
+
hidden layer of the Transformer.
|
| 798 |
+
|
| 799 |
+
Raises:
|
| 800 |
+
ValueError: A Tensor shape or parameter is invalid.
|
| 801 |
+
"""
|
| 802 |
+
if hidden_size % num_attention_heads != 0:
|
| 803 |
+
raise ValueError(
|
| 804 |
+
"The hidden size (%d) is not a multiple of the number of attention "
|
| 805 |
+
"heads (%d)" % (hidden_size, num_attention_heads))
|
| 806 |
+
|
| 807 |
+
attention_head_size = int(hidden_size / num_attention_heads)
|
| 808 |
+
input_shape = get_shape_list(input_tensor, expected_rank=3)
|
| 809 |
+
batch_size = input_shape[0]
|
| 810 |
+
seq_length = input_shape[1]
|
| 811 |
+
input_width = input_shape[2]
|
| 812 |
+
|
| 813 |
+
# The Transformer performs sum residuals on all layers so the input needs
|
| 814 |
+
# to be the same as the hidden size.
|
| 815 |
+
if input_width != hidden_size:
|
| 816 |
+
raise ValueError("The width of the input tensor (%d) != hidden size (%d)" %
|
| 817 |
+
(input_width, hidden_size))
|
| 818 |
+
|
| 819 |
+
# We keep the representation as a 2D tensor to avoid re-shaping it back and
|
| 820 |
+
# forth from a 3D tensor to a 2D tensor. Re-shapes are normally free on
|
| 821 |
+
# the GPU/CPU but may not be free on the TPU, so we want to minimize them to
|
| 822 |
+
# help the optimizer.
|
| 823 |
+
prev_output = reshape_to_matrix(input_tensor)
|
| 824 |
+
|
| 825 |
+
all_layer_outputs = []
|
| 826 |
+
for layer_idx in range(num_hidden_layers):
|
| 827 |
+
with tf.variable_scope("layer_%d" % layer_idx):
|
| 828 |
+
layer_input = prev_output
|
| 829 |
+
|
| 830 |
+
with tf.variable_scope("attention"):
|
| 831 |
+
attention_heads = []
|
| 832 |
+
with tf.variable_scope("self"):
|
| 833 |
+
attention_head = attention_layer(
|
| 834 |
+
from_tensor=layer_input,
|
| 835 |
+
to_tensor=layer_input,
|
| 836 |
+
attention_mask=attention_mask,
|
| 837 |
+
num_attention_heads=num_attention_heads,
|
| 838 |
+
size_per_head=attention_head_size,
|
| 839 |
+
attention_probs_dropout_prob=attention_probs_dropout_prob,
|
| 840 |
+
initializer_range=initializer_range,
|
| 841 |
+
do_return_2d_tensor=True,
|
| 842 |
+
batch_size=batch_size,
|
| 843 |
+
from_seq_length=seq_length,
|
| 844 |
+
to_seq_length=seq_length)
|
| 845 |
+
attention_heads.append(attention_head)
|
| 846 |
+
|
| 847 |
+
attention_output = None
|
| 848 |
+
if len(attention_heads) == 1:
|
| 849 |
+
attention_output = attention_heads[0]
|
| 850 |
+
else:
|
| 851 |
+
# In the case where we have other sequences, we just concatenate
|
| 852 |
+
# them to the self-attention head before the projection.
|
| 853 |
+
attention_output = tf.concat(attention_heads, axis=-1)
|
| 854 |
+
|
| 855 |
+
# Run a linear projection of `hidden_size` then add a residual
|
| 856 |
+
# with `layer_input`.
|
| 857 |
+
with tf.variable_scope("output"):
|
| 858 |
+
attention_output = tf.layers.dense(
|
| 859 |
+
attention_output,
|
| 860 |
+
hidden_size,
|
| 861 |
+
kernel_initializer=create_initializer(initializer_range))
|
| 862 |
+
attention_output = dropout(attention_output, hidden_dropout_prob)
|
| 863 |
+
attention_output = layer_norm(attention_output + layer_input)
|
| 864 |
+
|
| 865 |
+
# The activation is only applied to the "intermediate" hidden layer.
|
| 866 |
+
with tf.variable_scope("intermediate"):
|
| 867 |
+
intermediate_output = tf.layers.dense(
|
| 868 |
+
attention_output,
|
| 869 |
+
intermediate_size,
|
| 870 |
+
activation=intermediate_act_fn,
|
| 871 |
+
kernel_initializer=create_initializer(initializer_range))
|
| 872 |
+
|
| 873 |
+
# Down-project back to `hidden_size` then add the residual.
|
| 874 |
+
with tf.variable_scope("output"):
|
| 875 |
+
layer_output = tf.layers.dense(
|
| 876 |
+
intermediate_output,
|
| 877 |
+
hidden_size,
|
| 878 |
+
kernel_initializer=create_initializer(initializer_range))
|
| 879 |
+
layer_output = dropout(layer_output, hidden_dropout_prob)
|
| 880 |
+
layer_output = layer_norm(layer_output + attention_output)
|
| 881 |
+
prev_output = layer_output
|
| 882 |
+
all_layer_outputs.append(layer_output)
|
| 883 |
+
|
| 884 |
+
if do_return_all_layers:
|
| 885 |
+
final_outputs = []
|
| 886 |
+
for layer_output in all_layer_outputs:
|
| 887 |
+
final_output = reshape_from_matrix(layer_output, input_shape)
|
| 888 |
+
final_outputs.append(final_output)
|
| 889 |
+
return final_outputs
|
| 890 |
+
else:
|
| 891 |
+
final_output = reshape_from_matrix(prev_output, input_shape)
|
| 892 |
+
return final_output
|
| 893 |
+
|
| 894 |
+
|
| 895 |
+
def get_shape_list(tensor, expected_rank=None, name=None):
|
| 896 |
+
"""Returns a list of the shape of tensor, preferring static dimensions.
|
| 897 |
+
|
| 898 |
+
Args:
|
| 899 |
+
tensor: A tf.Tensor object to find the shape of.
|
| 900 |
+
expected_rank: (optional) int. The expected rank of `tensor`. If this is
|
| 901 |
+
specified and the `tensor` has a different rank, and exception will be
|
| 902 |
+
thrown.
|
| 903 |
+
name: Optional name of the tensor for the error message.
|
| 904 |
+
|
| 905 |
+
Returns:
|
| 906 |
+
A list of dimensions of the shape of tensor. All static dimensions will
|
| 907 |
+
be returned as python integers, and dynamic dimensions will be returned
|
| 908 |
+
as tf.Tensor scalars.
|
| 909 |
+
"""
|
| 910 |
+
if name is None:
|
| 911 |
+
name = tensor.name
|
| 912 |
+
|
| 913 |
+
if expected_rank is not None:
|
| 914 |
+
assert_rank(tensor, expected_rank, name)
|
| 915 |
+
|
| 916 |
+
shape = tensor.shape.as_list()
|
| 917 |
+
|
| 918 |
+
non_static_indexes = []
|
| 919 |
+
for (index, dim) in enumerate(shape):
|
| 920 |
+
if dim is None:
|
| 921 |
+
non_static_indexes.append(index)
|
| 922 |
+
|
| 923 |
+
if not non_static_indexes:
|
| 924 |
+
return shape
|
| 925 |
+
|
| 926 |
+
dyn_shape = tf.shape(tensor)
|
| 927 |
+
for index in non_static_indexes:
|
| 928 |
+
shape[index] = dyn_shape[index]
|
| 929 |
+
return shape
|
| 930 |
+
|
| 931 |
+
|
| 932 |
+
def reshape_to_matrix(input_tensor):
|
| 933 |
+
"""Reshapes a >= rank 2 tensor to a rank 2 tensor (i.e., a matrix)."""
|
| 934 |
+
ndims = input_tensor.shape.ndims
|
| 935 |
+
if ndims < 2:
|
| 936 |
+
raise ValueError("Input tensor must have at least rank 2. Shape = %s" %
|
| 937 |
+
(input_tensor.shape))
|
| 938 |
+
if ndims == 2:
|
| 939 |
+
return input_tensor
|
| 940 |
+
|
| 941 |
+
width = input_tensor.shape[-1]
|
| 942 |
+
output_tensor = tf.reshape(input_tensor, [-1, width])
|
| 943 |
+
return output_tensor
|
| 944 |
+
|
| 945 |
+
|
| 946 |
+
def reshape_from_matrix(output_tensor, orig_shape_list):
|
| 947 |
+
"""Reshapes a rank 2 tensor back to its original rank >= 2 tensor."""
|
| 948 |
+
if len(orig_shape_list) == 2:
|
| 949 |
+
return output_tensor
|
| 950 |
+
|
| 951 |
+
output_shape = get_shape_list(output_tensor)
|
| 952 |
+
|
| 953 |
+
orig_dims = orig_shape_list[0:-1]
|
| 954 |
+
width = output_shape[-1]
|
| 955 |
+
|
| 956 |
+
return tf.reshape(output_tensor, orig_dims + [width])
|
| 957 |
+
|
| 958 |
+
|
| 959 |
+
def assert_rank(tensor, expected_rank, name=None):
|
| 960 |
+
"""Raises an exception if the tensor rank is not of the expected rank.
|
| 961 |
+
|
| 962 |
+
Args:
|
| 963 |
+
tensor: A tf.Tensor to check the rank of.
|
| 964 |
+
expected_rank: Python integer or list of integers, expected rank.
|
| 965 |
+
name: Optional name of the tensor for the error message.
|
| 966 |
+
|
| 967 |
+
Raises:
|
| 968 |
+
ValueError: If the expected shape doesn't match the actual shape.
|
| 969 |
+
"""
|
| 970 |
+
if name is None:
|
| 971 |
+
name = tensor.name
|
| 972 |
+
|
| 973 |
+
expected_rank_dict = {}
|
| 974 |
+
if isinstance(expected_rank, six.integer_types):
|
| 975 |
+
expected_rank_dict[expected_rank] = True
|
| 976 |
+
else:
|
| 977 |
+
for x in expected_rank:
|
| 978 |
+
expected_rank_dict[x] = True
|
| 979 |
+
|
| 980 |
+
actual_rank = tensor.shape.ndims
|
| 981 |
+
if actual_rank not in expected_rank_dict:
|
| 982 |
+
scope_name = tf.get_variable_scope().name
|
| 983 |
+
raise ValueError(
|
| 984 |
+
"For the tensor `%s` in scope `%s`, the actual rank "
|
| 985 |
+
"`%d` (shape = %s) is not equal to the expected rank `%s`" %
|
| 986 |
+
(name, scope_name, actual_rank, str(tensor.shape), str(expected_rank)))
|
bert-master/bert-master/modeling_test.py
ADDED
|
@@ -0,0 +1,277 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# coding=utf-8
|
| 2 |
+
# Copyright 2018 The Google AI Language Team Authors.
|
| 3 |
+
#
|
| 4 |
+
# Licensed under the Apache License, Version 2.0 (the "License");
|
| 5 |
+
# you may not use this file except in compliance with the License.
|
| 6 |
+
# You may obtain a copy of the License at
|
| 7 |
+
#
|
| 8 |
+
# http://www.apache.org/licenses/LICENSE-2.0
|
| 9 |
+
#
|
| 10 |
+
# Unless required by applicable law or agreed to in writing, software
|
| 11 |
+
# distributed under the License is distributed on an "AS IS" BASIS,
|
| 12 |
+
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
| 13 |
+
# See the License for the specific language governing permissions and
|
| 14 |
+
# limitations under the License.
|
| 15 |
+
from __future__ import absolute_import
|
| 16 |
+
from __future__ import division
|
| 17 |
+
from __future__ import print_function
|
| 18 |
+
|
| 19 |
+
import collections
|
| 20 |
+
import json
|
| 21 |
+
import random
|
| 22 |
+
import re
|
| 23 |
+
|
| 24 |
+
import modeling
|
| 25 |
+
import six
|
| 26 |
+
import tensorflow as tf
|
| 27 |
+
|
| 28 |
+
|
| 29 |
+
class BertModelTest(tf.test.TestCase):
|
| 30 |
+
|
| 31 |
+
class BertModelTester(object):
|
| 32 |
+
|
| 33 |
+
def __init__(self,
|
| 34 |
+
parent,
|
| 35 |
+
batch_size=13,
|
| 36 |
+
seq_length=7,
|
| 37 |
+
is_training=True,
|
| 38 |
+
use_input_mask=True,
|
| 39 |
+
use_token_type_ids=True,
|
| 40 |
+
vocab_size=99,
|
| 41 |
+
hidden_size=32,
|
| 42 |
+
num_hidden_layers=5,
|
| 43 |
+
num_attention_heads=4,
|
| 44 |
+
intermediate_size=37,
|
| 45 |
+
hidden_act="gelu",
|
| 46 |
+
hidden_dropout_prob=0.1,
|
| 47 |
+
attention_probs_dropout_prob=0.1,
|
| 48 |
+
max_position_embeddings=512,
|
| 49 |
+
type_vocab_size=16,
|
| 50 |
+
initializer_range=0.02,
|
| 51 |
+
scope=None):
|
| 52 |
+
self.parent = parent
|
| 53 |
+
self.batch_size = batch_size
|
| 54 |
+
self.seq_length = seq_length
|
| 55 |
+
self.is_training = is_training
|
| 56 |
+
self.use_input_mask = use_input_mask
|
| 57 |
+
self.use_token_type_ids = use_token_type_ids
|
| 58 |
+
self.vocab_size = vocab_size
|
| 59 |
+
self.hidden_size = hidden_size
|
| 60 |
+
self.num_hidden_layers = num_hidden_layers
|
| 61 |
+
self.num_attention_heads = num_attention_heads
|
| 62 |
+
self.intermediate_size = intermediate_size
|
| 63 |
+
self.hidden_act = hidden_act
|
| 64 |
+
self.hidden_dropout_prob = hidden_dropout_prob
|
| 65 |
+
self.attention_probs_dropout_prob = attention_probs_dropout_prob
|
| 66 |
+
self.max_position_embeddings = max_position_embeddings
|
| 67 |
+
self.type_vocab_size = type_vocab_size
|
| 68 |
+
self.initializer_range = initializer_range
|
| 69 |
+
self.scope = scope
|
| 70 |
+
|
| 71 |
+
def create_model(self):
|
| 72 |
+
input_ids = BertModelTest.ids_tensor([self.batch_size, self.seq_length],
|
| 73 |
+
self.vocab_size)
|
| 74 |
+
|
| 75 |
+
input_mask = None
|
| 76 |
+
if self.use_input_mask:
|
| 77 |
+
input_mask = BertModelTest.ids_tensor(
|
| 78 |
+
[self.batch_size, self.seq_length], vocab_size=2)
|
| 79 |
+
|
| 80 |
+
token_type_ids = None
|
| 81 |
+
if self.use_token_type_ids:
|
| 82 |
+
token_type_ids = BertModelTest.ids_tensor(
|
| 83 |
+
[self.batch_size, self.seq_length], self.type_vocab_size)
|
| 84 |
+
|
| 85 |
+
config = modeling.BertConfig(
|
| 86 |
+
vocab_size=self.vocab_size,
|
| 87 |
+
hidden_size=self.hidden_size,
|
| 88 |
+
num_hidden_layers=self.num_hidden_layers,
|
| 89 |
+
num_attention_heads=self.num_attention_heads,
|
| 90 |
+
intermediate_size=self.intermediate_size,
|
| 91 |
+
hidden_act=self.hidden_act,
|
| 92 |
+
hidden_dropout_prob=self.hidden_dropout_prob,
|
| 93 |
+
attention_probs_dropout_prob=self.attention_probs_dropout_prob,
|
| 94 |
+
max_position_embeddings=self.max_position_embeddings,
|
| 95 |
+
type_vocab_size=self.type_vocab_size,
|
| 96 |
+
initializer_range=self.initializer_range)
|
| 97 |
+
|
| 98 |
+
model = modeling.BertModel(
|
| 99 |
+
config=config,
|
| 100 |
+
is_training=self.is_training,
|
| 101 |
+
input_ids=input_ids,
|
| 102 |
+
input_mask=input_mask,
|
| 103 |
+
token_type_ids=token_type_ids,
|
| 104 |
+
scope=self.scope)
|
| 105 |
+
|
| 106 |
+
outputs = {
|
| 107 |
+
"embedding_output": model.get_embedding_output(),
|
| 108 |
+
"sequence_output": model.get_sequence_output(),
|
| 109 |
+
"pooled_output": model.get_pooled_output(),
|
| 110 |
+
"all_encoder_layers": model.get_all_encoder_layers(),
|
| 111 |
+
}
|
| 112 |
+
return outputs
|
| 113 |
+
|
| 114 |
+
def check_output(self, result):
|
| 115 |
+
self.parent.assertAllEqual(
|
| 116 |
+
result["embedding_output"].shape,
|
| 117 |
+
[self.batch_size, self.seq_length, self.hidden_size])
|
| 118 |
+
|
| 119 |
+
self.parent.assertAllEqual(
|
| 120 |
+
result["sequence_output"].shape,
|
| 121 |
+
[self.batch_size, self.seq_length, self.hidden_size])
|
| 122 |
+
|
| 123 |
+
self.parent.assertAllEqual(result["pooled_output"].shape,
|
| 124 |
+
[self.batch_size, self.hidden_size])
|
| 125 |
+
|
| 126 |
+
def test_default(self):
|
| 127 |
+
self.run_tester(BertModelTest.BertModelTester(self))
|
| 128 |
+
|
| 129 |
+
def test_config_to_json_string(self):
|
| 130 |
+
config = modeling.BertConfig(vocab_size=99, hidden_size=37)
|
| 131 |
+
obj = json.loads(config.to_json_string())
|
| 132 |
+
self.assertEqual(obj["vocab_size"], 99)
|
| 133 |
+
self.assertEqual(obj["hidden_size"], 37)
|
| 134 |
+
|
| 135 |
+
def run_tester(self, tester):
|
| 136 |
+
with self.test_session() as sess:
|
| 137 |
+
ops = tester.create_model()
|
| 138 |
+
init_op = tf.group(tf.global_variables_initializer(),
|
| 139 |
+
tf.local_variables_initializer())
|
| 140 |
+
sess.run(init_op)
|
| 141 |
+
output_result = sess.run(ops)
|
| 142 |
+
tester.check_output(output_result)
|
| 143 |
+
|
| 144 |
+
self.assert_all_tensors_reachable(sess, [init_op, ops])
|
| 145 |
+
|
| 146 |
+
@classmethod
|
| 147 |
+
def ids_tensor(cls, shape, vocab_size, rng=None, name=None):
|
| 148 |
+
"""Creates a random int32 tensor of the shape within the vocab size."""
|
| 149 |
+
if rng is None:
|
| 150 |
+
rng = random.Random()
|
| 151 |
+
|
| 152 |
+
total_dims = 1
|
| 153 |
+
for dim in shape:
|
| 154 |
+
total_dims *= dim
|
| 155 |
+
|
| 156 |
+
values = []
|
| 157 |
+
for _ in range(total_dims):
|
| 158 |
+
values.append(rng.randint(0, vocab_size - 1))
|
| 159 |
+
|
| 160 |
+
return tf.constant(value=values, dtype=tf.int32, shape=shape, name=name)
|
| 161 |
+
|
| 162 |
+
def assert_all_tensors_reachable(self, sess, outputs):
|
| 163 |
+
"""Checks that all the tensors in the graph are reachable from outputs."""
|
| 164 |
+
graph = sess.graph
|
| 165 |
+
|
| 166 |
+
ignore_strings = [
|
| 167 |
+
"^.*/assert_less_equal/.*$",
|
| 168 |
+
"^.*/dilation_rate$",
|
| 169 |
+
"^.*/Tensordot/concat$",
|
| 170 |
+
"^.*/Tensordot/concat/axis$",
|
| 171 |
+
"^testing/.*$",
|
| 172 |
+
]
|
| 173 |
+
|
| 174 |
+
ignore_regexes = [re.compile(x) for x in ignore_strings]
|
| 175 |
+
|
| 176 |
+
unreachable = self.get_unreachable_ops(graph, outputs)
|
| 177 |
+
filtered_unreachable = []
|
| 178 |
+
for x in unreachable:
|
| 179 |
+
do_ignore = False
|
| 180 |
+
for r in ignore_regexes:
|
| 181 |
+
m = r.match(x.name)
|
| 182 |
+
if m is not None:
|
| 183 |
+
do_ignore = True
|
| 184 |
+
if do_ignore:
|
| 185 |
+
continue
|
| 186 |
+
filtered_unreachable.append(x)
|
| 187 |
+
unreachable = filtered_unreachable
|
| 188 |
+
|
| 189 |
+
self.assertEqual(
|
| 190 |
+
len(unreachable), 0, "The following ops are unreachable: %s" %
|
| 191 |
+
(" ".join([x.name for x in unreachable])))
|
| 192 |
+
|
| 193 |
+
@classmethod
|
| 194 |
+
def get_unreachable_ops(cls, graph, outputs):
|
| 195 |
+
"""Finds all of the tensors in graph that are unreachable from outputs."""
|
| 196 |
+
outputs = cls.flatten_recursive(outputs)
|
| 197 |
+
output_to_op = collections.defaultdict(list)
|
| 198 |
+
op_to_all = collections.defaultdict(list)
|
| 199 |
+
assign_out_to_in = collections.defaultdict(list)
|
| 200 |
+
|
| 201 |
+
for op in graph.get_operations():
|
| 202 |
+
for x in op.inputs:
|
| 203 |
+
op_to_all[op.name].append(x.name)
|
| 204 |
+
for y in op.outputs:
|
| 205 |
+
output_to_op[y.name].append(op.name)
|
| 206 |
+
op_to_all[op.name].append(y.name)
|
| 207 |
+
if str(op.type) == "Assign":
|
| 208 |
+
for y in op.outputs:
|
| 209 |
+
for x in op.inputs:
|
| 210 |
+
assign_out_to_in[y.name].append(x.name)
|
| 211 |
+
|
| 212 |
+
assign_groups = collections.defaultdict(list)
|
| 213 |
+
for out_name in assign_out_to_in.keys():
|
| 214 |
+
name_group = assign_out_to_in[out_name]
|
| 215 |
+
for n1 in name_group:
|
| 216 |
+
assign_groups[n1].append(out_name)
|
| 217 |
+
for n2 in name_group:
|
| 218 |
+
if n1 != n2:
|
| 219 |
+
assign_groups[n1].append(n2)
|
| 220 |
+
|
| 221 |
+
seen_tensors = {}
|
| 222 |
+
stack = [x.name for x in outputs]
|
| 223 |
+
while stack:
|
| 224 |
+
name = stack.pop()
|
| 225 |
+
if name in seen_tensors:
|
| 226 |
+
continue
|
| 227 |
+
seen_tensors[name] = True
|
| 228 |
+
|
| 229 |
+
if name in output_to_op:
|
| 230 |
+
for op_name in output_to_op[name]:
|
| 231 |
+
if op_name in op_to_all:
|
| 232 |
+
for input_name in op_to_all[op_name]:
|
| 233 |
+
if input_name not in stack:
|
| 234 |
+
stack.append(input_name)
|
| 235 |
+
|
| 236 |
+
expanded_names = []
|
| 237 |
+
if name in assign_groups:
|
| 238 |
+
for assign_name in assign_groups[name]:
|
| 239 |
+
expanded_names.append(assign_name)
|
| 240 |
+
|
| 241 |
+
for expanded_name in expanded_names:
|
| 242 |
+
if expanded_name not in stack:
|
| 243 |
+
stack.append(expanded_name)
|
| 244 |
+
|
| 245 |
+
unreachable_ops = []
|
| 246 |
+
for op in graph.get_operations():
|
| 247 |
+
is_unreachable = False
|
| 248 |
+
all_names = [x.name for x in op.inputs] + [x.name for x in op.outputs]
|
| 249 |
+
for name in all_names:
|
| 250 |
+
if name not in seen_tensors:
|
| 251 |
+
is_unreachable = True
|
| 252 |
+
if is_unreachable:
|
| 253 |
+
unreachable_ops.append(op)
|
| 254 |
+
return unreachable_ops
|
| 255 |
+
|
| 256 |
+
@classmethod
|
| 257 |
+
def flatten_recursive(cls, item):
|
| 258 |
+
"""Flattens (potentially nested) a tuple/dictionary/list to a list."""
|
| 259 |
+
output = []
|
| 260 |
+
if isinstance(item, list):
|
| 261 |
+
output.extend(item)
|
| 262 |
+
elif isinstance(item, tuple):
|
| 263 |
+
output.extend(list(item))
|
| 264 |
+
elif isinstance(item, dict):
|
| 265 |
+
for (_, v) in six.iteritems(item):
|
| 266 |
+
output.append(v)
|
| 267 |
+
else:
|
| 268 |
+
return [item]
|
| 269 |
+
|
| 270 |
+
flat_output = []
|
| 271 |
+
for x in output:
|
| 272 |
+
flat_output.extend(cls.flatten_recursive(x))
|
| 273 |
+
return flat_output
|
| 274 |
+
|
| 275 |
+
|
| 276 |
+
if __name__ == "__main__":
|
| 277 |
+
tf.test.main()
|
bert-master/bert-master/multilingual.md
ADDED
|
@@ -0,0 +1,303 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
## Models
|
| 2 |
+
|
| 3 |
+
There are two multilingual models currently available. We do not plan to release
|
| 4 |
+
more single-language models, but we may release `BERT-Large` versions of these
|
| 5 |
+
two in the future:
|
| 6 |
+
|
| 7 |
+
* **[`BERT-Base, Multilingual Cased (New, recommended)`](https://storage.googleapis.com/bert_models/2018_11_23/multi_cased_L-12_H-768_A-12.zip)**:
|
| 8 |
+
104 languages, 12-layer, 768-hidden, 12-heads, 110M parameters
|
| 9 |
+
* **[`BERT-Base, Multilingual Uncased (Orig, not recommended)`](https://storage.googleapis.com/bert_models/2018_11_03/multilingual_L-12_H-768_A-12.zip)**:
|
| 10 |
+
102 languages, 12-layer, 768-hidden, 12-heads, 110M parameters
|
| 11 |
+
* **[`BERT-Base, Chinese`](https://storage.googleapis.com/bert_models/2018_11_03/chinese_L-12_H-768_A-12.zip)**:
|
| 12 |
+
Chinese Simplified and Traditional, 12-layer, 768-hidden, 12-heads, 110M
|
| 13 |
+
parameters
|
| 14 |
+
|
| 15 |
+
**The `Multilingual Cased (New)` model also fixes normalization issues in many
|
| 16 |
+
languages, so it is recommended in languages with non-Latin alphabets (and is
|
| 17 |
+
often better for most languages with Latin alphabets). When using this model,
|
| 18 |
+
make sure to pass `--do_lower_case=false` to `run_pretraining.py` and other
|
| 19 |
+
scripts.**
|
| 20 |
+
|
| 21 |
+
See the [list of languages](#list-of-languages) that the Multilingual model
|
| 22 |
+
supports. The Multilingual model does include Chinese (and English), but if your
|
| 23 |
+
fine-tuning data is Chinese-only, then the Chinese model will likely produce
|
| 24 |
+
better results.
|
| 25 |
+
|
| 26 |
+
## Results
|
| 27 |
+
|
| 28 |
+
To evaluate these systems, we use the
|
| 29 |
+
[XNLI dataset](https://github.com/facebookresearch/XNLI) dataset, which is a
|
| 30 |
+
version of [MultiNLI](https://www.nyu.edu/projects/bowman/multinli/) where the
|
| 31 |
+
dev and test sets have been translated (by humans) into 15 languages. Note that
|
| 32 |
+
the training set was *machine* translated (we used the translations provided by
|
| 33 |
+
XNLI, not Google NMT). For clarity, we only report on 6 languages below:
|
| 34 |
+
|
| 35 |
+
<!-- mdformat off(no table) -->
|
| 36 |
+
|
| 37 |
+
| System | English | Chinese | Spanish | German | Arabic | Urdu |
|
| 38 |
+
| --------------------------------- | -------- | -------- | -------- | -------- | -------- | -------- |
|
| 39 |
+
| XNLI Baseline - Translate Train | 73.7 | 67.0 | 68.8 | 66.5 | 65.8 | 56.6 |
|
| 40 |
+
| XNLI Baseline - Translate Test | 73.7 | 68.3 | 70.7 | 68.7 | 66.8 | 59.3 |
|
| 41 |
+
| BERT - Translate Train Cased | **81.9** | **76.6** | **77.8** | **75.9** | **70.7** | 61.6 |
|
| 42 |
+
| BERT - Translate Train Uncased | 81.4 | 74.2 | 77.3 | 75.2 | 70.5 | 61.7 |
|
| 43 |
+
| BERT - Translate Test Uncased | 81.4 | 70.1 | 74.9 | 74.4 | 70.4 | **62.1** |
|
| 44 |
+
| BERT - Zero Shot Uncased | 81.4 | 63.8 | 74.3 | 70.5 | 62.1 | 58.3 |
|
| 45 |
+
|
| 46 |
+
<!-- mdformat on -->
|
| 47 |
+
|
| 48 |
+
The first two rows are baselines from the XNLI paper and the last three rows are
|
| 49 |
+
our results with BERT.
|
| 50 |
+
|
| 51 |
+
**Translate Train** means that the MultiNLI training set was machine translated
|
| 52 |
+
from English into the foreign language. So training and evaluation were both
|
| 53 |
+
done in the foreign language. Unfortunately, training was done on
|
| 54 |
+
machine-translated data, so it is impossible to quantify how much of the lower
|
| 55 |
+
accuracy (compared to English) is due to the quality of the machine translation
|
| 56 |
+
vs. the quality of the pre-trained model.
|
| 57 |
+
|
| 58 |
+
**Translate Test** means that the XNLI test set was machine translated from the
|
| 59 |
+
foreign language into English. So training and evaluation were both done on
|
| 60 |
+
English. However, test evaluation was done on machine-translated English, so the
|
| 61 |
+
accuracy depends on the quality of the machine translation system.
|
| 62 |
+
|
| 63 |
+
**Zero Shot** means that the Multilingual BERT system was fine-tuned on English
|
| 64 |
+
MultiNLI, and then evaluated on the foreign language XNLI test. In this case,
|
| 65 |
+
machine translation was not involved at all in either the pre-training or
|
| 66 |
+
fine-tuning.
|
| 67 |
+
|
| 68 |
+
Note that the English result is worse than the 84.2 MultiNLI baseline because
|
| 69 |
+
this training used Multilingual BERT rather than English-only BERT. This implies
|
| 70 |
+
that for high-resource languages, the Multilingual model is somewhat worse than
|
| 71 |
+
a single-language model. However, it is not feasible for us to train and
|
| 72 |
+
maintain dozens of single-language models. Therefore, if your goal is to maximize
|
| 73 |
+
performance with a language other than English or Chinese, you might find it
|
| 74 |
+
beneficial to run pre-training for additional steps starting from our
|
| 75 |
+
Multilingual model on data from your language of interest.
|
| 76 |
+
|
| 77 |
+
Here is a comparison of training Chinese models with the Multilingual
|
| 78 |
+
`BERT-Base` and Chinese-only `BERT-Base`:
|
| 79 |
+
|
| 80 |
+
System | Chinese
|
| 81 |
+
----------------------- | -------
|
| 82 |
+
XNLI Baseline | 67.0
|
| 83 |
+
BERT Multilingual Model | 74.2
|
| 84 |
+
BERT Chinese-only Model | 77.2
|
| 85 |
+
|
| 86 |
+
Similar to English, the single-language model does 3% better than the
|
| 87 |
+
Multilingual model.
|
| 88 |
+
|
| 89 |
+
## Fine-tuning Example
|
| 90 |
+
|
| 91 |
+
The multilingual model does **not** require any special consideration or API
|
| 92 |
+
changes. We did update the implementation of `BasicTokenizer` in
|
| 93 |
+
`tokenization.py` to support Chinese character tokenization, so please update if
|
| 94 |
+
you forked it. However, we did not change the tokenization API.
|
| 95 |
+
|
| 96 |
+
To test the new models, we did modify `run_classifier.py` to add support for the
|
| 97 |
+
[XNLI dataset](https://github.com/facebookresearch/XNLI). This is a 15-language
|
| 98 |
+
version of MultiNLI where the dev/test sets have been human-translated, and the
|
| 99 |
+
training set has been machine-translated.
|
| 100 |
+
|
| 101 |
+
To run the fine-tuning code, please download the
|
| 102 |
+
[XNLI dev/test set](https://www.nyu.edu/projects/bowman/xnli/XNLI-1.0.zip) and the
|
| 103 |
+
[XNLI machine-translated training set](https://www.nyu.edu/projects/bowman/xnli/XNLI-MT-1.0.zip)
|
| 104 |
+
and then unpack both .zip files into some directory `$XNLI_DIR`.
|
| 105 |
+
|
| 106 |
+
To run fine-tuning on XNLI. The language is hard-coded into `run_classifier.py`
|
| 107 |
+
(Chinese by default), so please modify `XnliProcessor` if you want to run on
|
| 108 |
+
another language.
|
| 109 |
+
|
| 110 |
+
This is a large dataset, so this will training will take a few hours on a GPU
|
| 111 |
+
(or about 30 minutes on a Cloud TPU). To run an experiment quickly for
|
| 112 |
+
debugging, just set `num_train_epochs` to a small value like `0.1`.
|
| 113 |
+
|
| 114 |
+
```shell
|
| 115 |
+
export BERT_BASE_DIR=/path/to/bert/chinese_L-12_H-768_A-12 # or multilingual_L-12_H-768_A-12
|
| 116 |
+
export XNLI_DIR=/path/to/xnli
|
| 117 |
+
|
| 118 |
+
python run_classifier.py \
|
| 119 |
+
--task_name=XNLI \
|
| 120 |
+
--do_train=true \
|
| 121 |
+
--do_eval=true \
|
| 122 |
+
--data_dir=$XNLI_DIR \
|
| 123 |
+
--vocab_file=$BERT_BASE_DIR/vocab.txt \
|
| 124 |
+
--bert_config_file=$BERT_BASE_DIR/bert_config.json \
|
| 125 |
+
--init_checkpoint=$BERT_BASE_DIR/bert_model.ckpt \
|
| 126 |
+
--max_seq_length=128 \
|
| 127 |
+
--train_batch_size=32 \
|
| 128 |
+
--learning_rate=5e-5 \
|
| 129 |
+
--num_train_epochs=2.0 \
|
| 130 |
+
--output_dir=/tmp/xnli_output/
|
| 131 |
+
```
|
| 132 |
+
|
| 133 |
+
With the Chinese-only model, the results should look something like this:
|
| 134 |
+
|
| 135 |
+
```
|
| 136 |
+
***** Eval results *****
|
| 137 |
+
eval_accuracy = 0.774116
|
| 138 |
+
eval_loss = 0.83554
|
| 139 |
+
global_step = 24543
|
| 140 |
+
loss = 0.74603
|
| 141 |
+
```
|
| 142 |
+
|
| 143 |
+
## Details
|
| 144 |
+
|
| 145 |
+
### Data Source and Sampling
|
| 146 |
+
|
| 147 |
+
The languages chosen were the
|
| 148 |
+
[top 100 languages with the largest Wikipedias](https://meta.wikimedia.org/wiki/List_of_Wikipedias).
|
| 149 |
+
The entire Wikipedia dump for each language (excluding user and talk pages) was
|
| 150 |
+
taken as the training data for each language
|
| 151 |
+
|
| 152 |
+
However, the size of the Wikipedia for a given language varies greatly, and
|
| 153 |
+
therefore low-resource languages may be "under-represented" in terms of the
|
| 154 |
+
neural network model (under the assumption that languages are "competing" for
|
| 155 |
+
limited model capacity to some extent). At the same time, we also don't want
|
| 156 |
+
to overfit the model by performing thousands of epochs over a tiny Wikipedia
|
| 157 |
+
for a particular language.
|
| 158 |
+
|
| 159 |
+
To balance these two factors, we performed exponentially smoothed weighting of
|
| 160 |
+
the data during pre-training data creation (and WordPiece vocab creation). In
|
| 161 |
+
other words, let's say that the probability of a language is *P(L)*, e.g.,
|
| 162 |
+
*P(English) = 0.21* means that after concatenating all of the Wikipedias
|
| 163 |
+
together, 21% of our data is English. We exponentiate each probability by some
|
| 164 |
+
factor *S* and then re-normalize, and sample from that distribution. In our case
|
| 165 |
+
we use *S=0.7*. So, high-resource languages like English will be under-sampled,
|
| 166 |
+
and low-resource languages like Icelandic will be over-sampled. E.g., in the
|
| 167 |
+
original distribution English would be sampled 1000x more than Icelandic, but
|
| 168 |
+
after smoothing it's only sampled 100x more.
|
| 169 |
+
|
| 170 |
+
### Tokenization
|
| 171 |
+
|
| 172 |
+
For tokenization, we use a 110k shared WordPiece vocabulary. The word counts are
|
| 173 |
+
weighted the same way as the data, so low-resource languages are upweighted by
|
| 174 |
+
some factor. We intentionally do *not* use any marker to denote the input
|
| 175 |
+
language (so that zero-shot training can work).
|
| 176 |
+
|
| 177 |
+
Because Chinese (and Japanese Kanji and Korean Hanja) does not have whitespace
|
| 178 |
+
characters, we add spaces around every character in the
|
| 179 |
+
[CJK Unicode range](https://en.wikipedia.org/wiki/CJK_Unified_Ideographs_\(Unicode_block\))
|
| 180 |
+
before applying WordPiece. This means that Chinese is effectively
|
| 181 |
+
character-tokenized. Note that the CJK Unicode block only includes
|
| 182 |
+
Chinese-origin characters and does *not* include Hangul Korean or
|
| 183 |
+
Katakana/Hiragana Japanese, which are tokenized with whitespace+WordPiece like
|
| 184 |
+
all other languages.
|
| 185 |
+
|
| 186 |
+
For all other languages, we apply the
|
| 187 |
+
[same recipe as English](https://github.com/google-research/bert#tokenization):
|
| 188 |
+
(a) lower casing+accent removal, (b) punctuation splitting, (c) whitespace
|
| 189 |
+
tokenization. We understand that accent markers have substantial meaning in some
|
| 190 |
+
languages, but felt that the benefits of reducing the effective vocabulary make
|
| 191 |
+
up for this. Generally the strong contextual models of BERT should make up for
|
| 192 |
+
any ambiguity introduced by stripping accent markers.
|
| 193 |
+
|
| 194 |
+
### List of Languages
|
| 195 |
+
|
| 196 |
+
The multilingual model supports the following languages. These languages were
|
| 197 |
+
chosen because they are the top 100 languages with the largest Wikipedias:
|
| 198 |
+
|
| 199 |
+
* Afrikaans
|
| 200 |
+
* Albanian
|
| 201 |
+
* Arabic
|
| 202 |
+
* Aragonese
|
| 203 |
+
* Armenian
|
| 204 |
+
* Asturian
|
| 205 |
+
* Azerbaijani
|
| 206 |
+
* Bashkir
|
| 207 |
+
* Basque
|
| 208 |
+
* Bavarian
|
| 209 |
+
* Belarusian
|
| 210 |
+
* Bengali
|
| 211 |
+
* Bishnupriya Manipuri
|
| 212 |
+
* Bosnian
|
| 213 |
+
* Breton
|
| 214 |
+
* Bulgarian
|
| 215 |
+
* Burmese
|
| 216 |
+
* Catalan
|
| 217 |
+
* Cebuano
|
| 218 |
+
* Chechen
|
| 219 |
+
* Chinese (Simplified)
|
| 220 |
+
* Chinese (Traditional)
|
| 221 |
+
* Chuvash
|
| 222 |
+
* Croatian
|
| 223 |
+
* Czech
|
| 224 |
+
* Danish
|
| 225 |
+
* Dutch
|
| 226 |
+
* English
|
| 227 |
+
* Estonian
|
| 228 |
+
* Finnish
|
| 229 |
+
* French
|
| 230 |
+
* Galician
|
| 231 |
+
* Georgian
|
| 232 |
+
* German
|
| 233 |
+
* Greek
|
| 234 |
+
* Gujarati
|
| 235 |
+
* Haitian
|
| 236 |
+
* Hebrew
|
| 237 |
+
* Hindi
|
| 238 |
+
* Hungarian
|
| 239 |
+
* Icelandic
|
| 240 |
+
* Ido
|
| 241 |
+
* Indonesian
|
| 242 |
+
* Irish
|
| 243 |
+
* Italian
|
| 244 |
+
* Japanese
|
| 245 |
+
* Javanese
|
| 246 |
+
* Kannada
|
| 247 |
+
* Kazakh
|
| 248 |
+
* Kirghiz
|
| 249 |
+
* Korean
|
| 250 |
+
* Latin
|
| 251 |
+
* Latvian
|
| 252 |
+
* Lithuanian
|
| 253 |
+
* Lombard
|
| 254 |
+
* Low Saxon
|
| 255 |
+
* Luxembourgish
|
| 256 |
+
* Macedonian
|
| 257 |
+
* Malagasy
|
| 258 |
+
* Malay
|
| 259 |
+
* Malayalam
|
| 260 |
+
* Marathi
|
| 261 |
+
* Minangkabau
|
| 262 |
+
* Nepali
|
| 263 |
+
* Newar
|
| 264 |
+
* Norwegian (Bokmal)
|
| 265 |
+
* Norwegian (Nynorsk)
|
| 266 |
+
* Occitan
|
| 267 |
+
* Persian (Farsi)
|
| 268 |
+
* Piedmontese
|
| 269 |
+
* Polish
|
| 270 |
+
* Portuguese
|
| 271 |
+
* Punjabi
|
| 272 |
+
* Romanian
|
| 273 |
+
* Russian
|
| 274 |
+
* Scots
|
| 275 |
+
* Serbian
|
| 276 |
+
* Serbo-Croatian
|
| 277 |
+
* Sicilian
|
| 278 |
+
* Slovak
|
| 279 |
+
* Slovenian
|
| 280 |
+
* South Azerbaijani
|
| 281 |
+
* Spanish
|
| 282 |
+
* Sundanese
|
| 283 |
+
* Swahili
|
| 284 |
+
* Swedish
|
| 285 |
+
* Tagalog
|
| 286 |
+
* Tajik
|
| 287 |
+
* Tamil
|
| 288 |
+
* Tatar
|
| 289 |
+
* Telugu
|
| 290 |
+
* Turkish
|
| 291 |
+
* Ukrainian
|
| 292 |
+
* Urdu
|
| 293 |
+
* Uzbek
|
| 294 |
+
* Vietnamese
|
| 295 |
+
* Volapük
|
| 296 |
+
* Waray-Waray
|
| 297 |
+
* Welsh
|
| 298 |
+
* West Frisian
|
| 299 |
+
* Western Punjabi
|
| 300 |
+
* Yoruba
|
| 301 |
+
|
| 302 |
+
The **Multilingual Cased (New)** release contains additionally **Thai** and
|
| 303 |
+
**Mongolian**, which were not included in the original release.
|
bert-master/bert-master/optimization.py
ADDED
|
@@ -0,0 +1,174 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# coding=utf-8
|
| 2 |
+
# Copyright 2018 The Google AI Language Team Authors.
|
| 3 |
+
#
|
| 4 |
+
# Licensed under the Apache License, Version 2.0 (the "License");
|
| 5 |
+
# you may not use this file except in compliance with the License.
|
| 6 |
+
# You may obtain a copy of the License at
|
| 7 |
+
#
|
| 8 |
+
# http://www.apache.org/licenses/LICENSE-2.0
|
| 9 |
+
#
|
| 10 |
+
# Unless required by applicable law or agreed to in writing, software
|
| 11 |
+
# distributed under the License is distributed on an "AS IS" BASIS,
|
| 12 |
+
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
| 13 |
+
# See the License for the specific language governing permissions and
|
| 14 |
+
# limitations under the License.
|
| 15 |
+
"""Functions and classes related to optimization (weight updates)."""
|
| 16 |
+
|
| 17 |
+
from __future__ import absolute_import
|
| 18 |
+
from __future__ import division
|
| 19 |
+
from __future__ import print_function
|
| 20 |
+
|
| 21 |
+
import re
|
| 22 |
+
import tensorflow as tf
|
| 23 |
+
|
| 24 |
+
|
| 25 |
+
def create_optimizer(loss, init_lr, num_train_steps, num_warmup_steps, use_tpu):
|
| 26 |
+
"""Creates an optimizer training op."""
|
| 27 |
+
global_step = tf.train.get_or_create_global_step()
|
| 28 |
+
|
| 29 |
+
learning_rate = tf.constant(value=init_lr, shape=[], dtype=tf.float32)
|
| 30 |
+
|
| 31 |
+
# Implements linear decay of the learning rate.
|
| 32 |
+
learning_rate = tf.train.polynomial_decay(
|
| 33 |
+
learning_rate,
|
| 34 |
+
global_step,
|
| 35 |
+
num_train_steps,
|
| 36 |
+
end_learning_rate=0.0,
|
| 37 |
+
power=1.0,
|
| 38 |
+
cycle=False)
|
| 39 |
+
|
| 40 |
+
# Implements linear warmup. I.e., if global_step < num_warmup_steps, the
|
| 41 |
+
# learning rate will be `global_step/num_warmup_steps * init_lr`.
|
| 42 |
+
if num_warmup_steps:
|
| 43 |
+
global_steps_int = tf.cast(global_step, tf.int32)
|
| 44 |
+
warmup_steps_int = tf.constant(num_warmup_steps, dtype=tf.int32)
|
| 45 |
+
|
| 46 |
+
global_steps_float = tf.cast(global_steps_int, tf.float32)
|
| 47 |
+
warmup_steps_float = tf.cast(warmup_steps_int, tf.float32)
|
| 48 |
+
|
| 49 |
+
warmup_percent_done = global_steps_float / warmup_steps_float
|
| 50 |
+
warmup_learning_rate = init_lr * warmup_percent_done
|
| 51 |
+
|
| 52 |
+
is_warmup = tf.cast(global_steps_int < warmup_steps_int, tf.float32)
|
| 53 |
+
learning_rate = (
|
| 54 |
+
(1.0 - is_warmup) * learning_rate + is_warmup * warmup_learning_rate)
|
| 55 |
+
|
| 56 |
+
# It is recommended that you use this optimizer for fine tuning, since this
|
| 57 |
+
# is how the model was trained (note that the Adam m/v variables are NOT
|
| 58 |
+
# loaded from init_checkpoint.)
|
| 59 |
+
optimizer = AdamWeightDecayOptimizer(
|
| 60 |
+
learning_rate=learning_rate,
|
| 61 |
+
weight_decay_rate=0.01,
|
| 62 |
+
beta_1=0.9,
|
| 63 |
+
beta_2=0.999,
|
| 64 |
+
epsilon=1e-6,
|
| 65 |
+
exclude_from_weight_decay=["LayerNorm", "layer_norm", "bias"])
|
| 66 |
+
|
| 67 |
+
if use_tpu:
|
| 68 |
+
optimizer = tf.contrib.tpu.CrossShardOptimizer(optimizer)
|
| 69 |
+
|
| 70 |
+
tvars = tf.trainable_variables()
|
| 71 |
+
grads = tf.gradients(loss, tvars)
|
| 72 |
+
|
| 73 |
+
# This is how the model was pre-trained.
|
| 74 |
+
(grads, _) = tf.clip_by_global_norm(grads, clip_norm=1.0)
|
| 75 |
+
|
| 76 |
+
train_op = optimizer.apply_gradients(
|
| 77 |
+
zip(grads, tvars), global_step=global_step)
|
| 78 |
+
|
| 79 |
+
# Normally the global step update is done inside of `apply_gradients`.
|
| 80 |
+
# However, `AdamWeightDecayOptimizer` doesn't do this. But if you use
|
| 81 |
+
# a different optimizer, you should probably take this line out.
|
| 82 |
+
new_global_step = global_step + 1
|
| 83 |
+
train_op = tf.group(train_op, [global_step.assign(new_global_step)])
|
| 84 |
+
return train_op
|
| 85 |
+
|
| 86 |
+
|
| 87 |
+
class AdamWeightDecayOptimizer(tf.train.Optimizer):
|
| 88 |
+
"""A basic Adam optimizer that includes "correct" L2 weight decay."""
|
| 89 |
+
|
| 90 |
+
def __init__(self,
|
| 91 |
+
learning_rate,
|
| 92 |
+
weight_decay_rate=0.0,
|
| 93 |
+
beta_1=0.9,
|
| 94 |
+
beta_2=0.999,
|
| 95 |
+
epsilon=1e-6,
|
| 96 |
+
exclude_from_weight_decay=None,
|
| 97 |
+
name="AdamWeightDecayOptimizer"):
|
| 98 |
+
"""Constructs a AdamWeightDecayOptimizer."""
|
| 99 |
+
super(AdamWeightDecayOptimizer, self).__init__(False, name)
|
| 100 |
+
|
| 101 |
+
self.learning_rate = learning_rate
|
| 102 |
+
self.weight_decay_rate = weight_decay_rate
|
| 103 |
+
self.beta_1 = beta_1
|
| 104 |
+
self.beta_2 = beta_2
|
| 105 |
+
self.epsilon = epsilon
|
| 106 |
+
self.exclude_from_weight_decay = exclude_from_weight_decay
|
| 107 |
+
|
| 108 |
+
def apply_gradients(self, grads_and_vars, global_step=None, name=None):
|
| 109 |
+
"""See base class."""
|
| 110 |
+
assignments = []
|
| 111 |
+
for (grad, param) in grads_and_vars:
|
| 112 |
+
if grad is None or param is None:
|
| 113 |
+
continue
|
| 114 |
+
|
| 115 |
+
param_name = self._get_variable_name(param.name)
|
| 116 |
+
|
| 117 |
+
m = tf.get_variable(
|
| 118 |
+
name=param_name + "/adam_m",
|
| 119 |
+
shape=param.shape.as_list(),
|
| 120 |
+
dtype=tf.float32,
|
| 121 |
+
trainable=False,
|
| 122 |
+
initializer=tf.zeros_initializer())
|
| 123 |
+
v = tf.get_variable(
|
| 124 |
+
name=param_name + "/adam_v",
|
| 125 |
+
shape=param.shape.as_list(),
|
| 126 |
+
dtype=tf.float32,
|
| 127 |
+
trainable=False,
|
| 128 |
+
initializer=tf.zeros_initializer())
|
| 129 |
+
|
| 130 |
+
# Standard Adam update.
|
| 131 |
+
next_m = (
|
| 132 |
+
tf.multiply(self.beta_1, m) + tf.multiply(1.0 - self.beta_1, grad))
|
| 133 |
+
next_v = (
|
| 134 |
+
tf.multiply(self.beta_2, v) + tf.multiply(1.0 - self.beta_2,
|
| 135 |
+
tf.square(grad)))
|
| 136 |
+
|
| 137 |
+
update = next_m / (tf.sqrt(next_v) + self.epsilon)
|
| 138 |
+
|
| 139 |
+
# Just adding the square of the weights to the loss function is *not*
|
| 140 |
+
# the correct way of using L2 regularization/weight decay with Adam,
|
| 141 |
+
# since that will interact with the m and v parameters in strange ways.
|
| 142 |
+
#
|
| 143 |
+
# Instead we want ot decay the weights in a manner that doesn't interact
|
| 144 |
+
# with the m/v parameters. This is equivalent to adding the square
|
| 145 |
+
# of the weights to the loss with plain (non-momentum) SGD.
|
| 146 |
+
if self._do_use_weight_decay(param_name):
|
| 147 |
+
update += self.weight_decay_rate * param
|
| 148 |
+
|
| 149 |
+
update_with_lr = self.learning_rate * update
|
| 150 |
+
|
| 151 |
+
next_param = param - update_with_lr
|
| 152 |
+
|
| 153 |
+
assignments.extend(
|
| 154 |
+
[param.assign(next_param),
|
| 155 |
+
m.assign(next_m),
|
| 156 |
+
v.assign(next_v)])
|
| 157 |
+
return tf.group(*assignments, name=name)
|
| 158 |
+
|
| 159 |
+
def _do_use_weight_decay(self, param_name):
|
| 160 |
+
"""Whether to use L2 weight decay for `param_name`."""
|
| 161 |
+
if not self.weight_decay_rate:
|
| 162 |
+
return False
|
| 163 |
+
if self.exclude_from_weight_decay:
|
| 164 |
+
for r in self.exclude_from_weight_decay:
|
| 165 |
+
if re.search(r, param_name) is not None:
|
| 166 |
+
return False
|
| 167 |
+
return True
|
| 168 |
+
|
| 169 |
+
def _get_variable_name(self, param_name):
|
| 170 |
+
"""Get the variable name from the tensor name."""
|
| 171 |
+
m = re.match("^(.*):\\d+$", param_name)
|
| 172 |
+
if m is not None:
|
| 173 |
+
param_name = m.group(1)
|
| 174 |
+
return param_name
|
bert-master/bert-master/optimization_test.py
ADDED
|
@@ -0,0 +1,48 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# coding=utf-8
|
| 2 |
+
# Copyright 2018 The Google AI Language Team Authors.
|
| 3 |
+
#
|
| 4 |
+
# Licensed under the Apache License, Version 2.0 (the "License");
|
| 5 |
+
# you may not use this file except in compliance with the License.
|
| 6 |
+
# You may obtain a copy of the License at
|
| 7 |
+
#
|
| 8 |
+
# http://www.apache.org/licenses/LICENSE-2.0
|
| 9 |
+
#
|
| 10 |
+
# Unless required by applicable law or agreed to in writing, software
|
| 11 |
+
# distributed under the License is distributed on an "AS IS" BASIS,
|
| 12 |
+
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
| 13 |
+
# See the License for the specific language governing permissions and
|
| 14 |
+
# limitations under the License.
|
| 15 |
+
from __future__ import absolute_import
|
| 16 |
+
from __future__ import division
|
| 17 |
+
from __future__ import print_function
|
| 18 |
+
|
| 19 |
+
import optimization
|
| 20 |
+
import tensorflow as tf
|
| 21 |
+
|
| 22 |
+
|
| 23 |
+
class OptimizationTest(tf.test.TestCase):
|
| 24 |
+
|
| 25 |
+
def test_adam(self):
|
| 26 |
+
with self.test_session() as sess:
|
| 27 |
+
w = tf.get_variable(
|
| 28 |
+
"w",
|
| 29 |
+
shape=[3],
|
| 30 |
+
initializer=tf.constant_initializer([0.1, -0.2, -0.1]))
|
| 31 |
+
x = tf.constant([0.4, 0.2, -0.5])
|
| 32 |
+
loss = tf.reduce_mean(tf.square(x - w))
|
| 33 |
+
tvars = tf.trainable_variables()
|
| 34 |
+
grads = tf.gradients(loss, tvars)
|
| 35 |
+
global_step = tf.train.get_or_create_global_step()
|
| 36 |
+
optimizer = optimization.AdamWeightDecayOptimizer(learning_rate=0.2)
|
| 37 |
+
train_op = optimizer.apply_gradients(zip(grads, tvars), global_step)
|
| 38 |
+
init_op = tf.group(tf.global_variables_initializer(),
|
| 39 |
+
tf.local_variables_initializer())
|
| 40 |
+
sess.run(init_op)
|
| 41 |
+
for _ in range(100):
|
| 42 |
+
sess.run(train_op)
|
| 43 |
+
w_np = sess.run(w)
|
| 44 |
+
self.assertAllClose(w_np.flat, [0.4, 0.2, -0.5], rtol=1e-2, atol=1e-2)
|
| 45 |
+
|
| 46 |
+
|
| 47 |
+
if __name__ == "__main__":
|
| 48 |
+
tf.test.main()
|
bert-master/bert-master/predicting_movie_reviews_with_bert_on_tf_hub.ipynb
ADDED
|
@@ -0,0 +1,1231 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"nbformat": 4,
|
| 3 |
+
"nbformat_minor": 0,
|
| 4 |
+
"metadata": {
|
| 5 |
+
"colab": {
|
| 6 |
+
"name": "Predicting Movie Reviews with BERT on TF Hub.ipynb",
|
| 7 |
+
"version": "0.3.2",
|
| 8 |
+
"provenance": [],
|
| 9 |
+
"collapsed_sections": []
|
| 10 |
+
},
|
| 11 |
+
"kernelspec": {
|
| 12 |
+
"name": "python3",
|
| 13 |
+
"display_name": "Python 3"
|
| 14 |
+
},
|
| 15 |
+
"accelerator": "GPU"
|
| 16 |
+
},
|
| 17 |
+
"cells": [
|
| 18 |
+
{
|
| 19 |
+
"metadata": {
|
| 20 |
+
"id": "j0a4mTk9o1Qg",
|
| 21 |
+
"colab_type": "code",
|
| 22 |
+
"colab": {}
|
| 23 |
+
},
|
| 24 |
+
"cell_type": "code",
|
| 25 |
+
"source": [
|
| 26 |
+
"# Copyright 2019 Google Inc.\n",
|
| 27 |
+
"\n",
|
| 28 |
+
"# Licensed under the Apache License, Version 2.0 (the \"License\");\n",
|
| 29 |
+
"# you may not use this file except in compliance with the License.\n",
|
| 30 |
+
"# You may obtain a copy of the License at\n",
|
| 31 |
+
"\n",
|
| 32 |
+
"# http://www.apache.org/licenses/LICENSE-2.0\n",
|
| 33 |
+
"\n",
|
| 34 |
+
"# Unless required by applicable law or agreed to in writing, software\n",
|
| 35 |
+
"# distributed under the License is distributed on an \"AS IS\" BASIS,\n",
|
| 36 |
+
"# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.\n",
|
| 37 |
+
"# See the License for the specific language governing permissions and\n",
|
| 38 |
+
"# limitations under the License."
|
| 39 |
+
],
|
| 40 |
+
"execution_count": 0,
|
| 41 |
+
"outputs": []
|
| 42 |
+
},
|
| 43 |
+
{
|
| 44 |
+
"metadata": {
|
| 45 |
+
"id": "dCpvgG0vwXAZ",
|
| 46 |
+
"colab_type": "text"
|
| 47 |
+
},
|
| 48 |
+
"cell_type": "markdown",
|
| 49 |
+
"source": [
|
| 50 |
+
"#Predicting Movie Review Sentiment with BERT on TF Hub"
|
| 51 |
+
]
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"metadata": {
|
| 55 |
+
"id": "xiYrZKaHwV81",
|
| 56 |
+
"colab_type": "text"
|
| 57 |
+
},
|
| 58 |
+
"cell_type": "markdown",
|
| 59 |
+
"source": [
|
| 60 |
+
"If you’ve been following Natural Language Processing over the past year, you’ve probably heard of BERT: Bidirectional Encoder Representations from Transformers. It’s a neural network architecture designed by Google researchers that’s totally transformed what’s state-of-the-art for NLP tasks, like text classification, translation, summarization, and question answering.\n",
|
| 61 |
+
"\n",
|
| 62 |
+
"Now that BERT's been added to [TF Hub](https://www.tensorflow.org/hub) as a loadable module, it's easy(ish) to add into existing Tensorflow text pipelines. In an existing pipeline, BERT can replace text embedding layers like ELMO and GloVE. Alternatively, [finetuning](http://wiki.fast.ai/index.php/Fine_tuning) BERT can provide both an accuracy boost and faster training time in many cases.\n",
|
| 63 |
+
"\n",
|
| 64 |
+
"Here, we'll train a model to predict whether an IMDB movie review is positive or negative using BERT in Tensorflow with tf hub. Some code was adapted from [this colab notebook](https://colab.sandbox.google.com/github/tensorflow/tpu/blob/master/tools/colab/bert_finetuning_with_cloud_tpus.ipynb). Let's get started!"
|
| 65 |
+
]
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"metadata": {
|
| 69 |
+
"id": "hsZvic2YxnTz",
|
| 70 |
+
"colab_type": "code",
|
| 71 |
+
"colab": {}
|
| 72 |
+
},
|
| 73 |
+
"cell_type": "code",
|
| 74 |
+
"source": [
|
| 75 |
+
"from sklearn.model_selection import train_test_split\n",
|
| 76 |
+
"import pandas as pd\n",
|
| 77 |
+
"import tensorflow as tf\n",
|
| 78 |
+
"import tensorflow_hub as hub\n",
|
| 79 |
+
"from datetime import datetime"
|
| 80 |
+
],
|
| 81 |
+
"execution_count": 0,
|
| 82 |
+
"outputs": []
|
| 83 |
+
},
|
| 84 |
+
{
|
| 85 |
+
"metadata": {
|
| 86 |
+
"id": "cp5wfXDx5SPH",
|
| 87 |
+
"colab_type": "text"
|
| 88 |
+
},
|
| 89 |
+
"cell_type": "markdown",
|
| 90 |
+
"source": [
|
| 91 |
+
"In addition to the standard libraries we imported above, we'll need to install BERT's python package."
|
| 92 |
+
]
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"metadata": {
|
| 96 |
+
"id": "jviywGyWyKsA",
|
| 97 |
+
"colab_type": "code",
|
| 98 |
+
"outputId": "166f3005-d219-404f-b201-2a0b75480360",
|
| 99 |
+
"colab": {
|
| 100 |
+
"base_uri": "https://localhost:8080/",
|
| 101 |
+
"height": 51
|
| 102 |
+
}
|
| 103 |
+
},
|
| 104 |
+
"cell_type": "code",
|
| 105 |
+
"source": [
|
| 106 |
+
"!pip install bert-tensorflow"
|
| 107 |
+
],
|
| 108 |
+
"execution_count": 38,
|
| 109 |
+
"outputs": [
|
| 110 |
+
{
|
| 111 |
+
"output_type": "stream",
|
| 112 |
+
"text": [
|
| 113 |
+
"Requirement already satisfied: bert-tensorflow in /usr/local/lib/python3.6/dist-packages (1.0.1)\n",
|
| 114 |
+
"Requirement already satisfied: six in /usr/local/lib/python3.6/dist-packages (from bert-tensorflow) (1.11.0)\n"
|
| 115 |
+
],
|
| 116 |
+
"name": "stdout"
|
| 117 |
+
}
|
| 118 |
+
]
|
| 119 |
+
},
|
| 120 |
+
{
|
| 121 |
+
"metadata": {
|
| 122 |
+
"id": "hhbGEfwgdEtw",
|
| 123 |
+
"colab_type": "code",
|
| 124 |
+
"colab": {}
|
| 125 |
+
},
|
| 126 |
+
"cell_type": "code",
|
| 127 |
+
"source": [
|
| 128 |
+
"import bert\n",
|
| 129 |
+
"from bert import run_classifier\n",
|
| 130 |
+
"from bert import optimization\n",
|
| 131 |
+
"from bert import tokenization"
|
| 132 |
+
],
|
| 133 |
+
"execution_count": 0,
|
| 134 |
+
"outputs": []
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"metadata": {
|
| 138 |
+
"id": "KVB3eOcjxxm1",
|
| 139 |
+
"colab_type": "text"
|
| 140 |
+
},
|
| 141 |
+
"cell_type": "markdown",
|
| 142 |
+
"source": [
|
| 143 |
+
"Below, we'll set an output directory location to store our model output and checkpoints. This can be a local directory, in which case you'd set OUTPUT_DIR to the name of the directory you'd like to create. If you're running this code in Google's hosted Colab, the directory won't persist after the Colab session ends.\n",
|
| 144 |
+
"\n",
|
| 145 |
+
"Alternatively, if you're a GCP user, you can store output in a GCP bucket. To do that, set a directory name in OUTPUT_DIR and the name of the GCP bucket in the BUCKET field.\n",
|
| 146 |
+
"\n",
|
| 147 |
+
"Set DO_DELETE to rewrite the OUTPUT_DIR if it exists. Otherwise, Tensorflow will load existing model checkpoints from that directory (if they exist)."
|
| 148 |
+
]
|
| 149 |
+
},
|
| 150 |
+
{
|
| 151 |
+
"metadata": {
|
| 152 |
+
"id": "US_EAnICvP7f",
|
| 153 |
+
"colab_type": "code",
|
| 154 |
+
"outputId": "7780a032-31d4-4794-e6aa-664a5d2ae7dd",
|
| 155 |
+
"cellView": "form",
|
| 156 |
+
"colab": {
|
| 157 |
+
"base_uri": "https://localhost:8080/",
|
| 158 |
+
"height": 34
|
| 159 |
+
}
|
| 160 |
+
},
|
| 161 |
+
"cell_type": "code",
|
| 162 |
+
"source": [
|
| 163 |
+
"# Set the output directory for saving model file\n",
|
| 164 |
+
"# Optionally, set a GCP bucket location\n",
|
| 165 |
+
"\n",
|
| 166 |
+
"OUTPUT_DIR = 'OUTPUT_DIR_NAME'#@param {type:\"string\"}\n",
|
| 167 |
+
"#@markdown Whether or not to clear/delete the directory and create a new one\n",
|
| 168 |
+
"DO_DELETE = False #@param {type:\"boolean\"}\n",
|
| 169 |
+
"#@markdown Set USE_BUCKET and BUCKET if you want to (optionally) store model output on GCP bucket.\n",
|
| 170 |
+
"USE_BUCKET = True #@param {type:\"boolean\"}\n",
|
| 171 |
+
"BUCKET = 'BUCKET_NAME' #@param {type:\"string\"}\n",
|
| 172 |
+
"\n",
|
| 173 |
+
"if USE_BUCKET:\n",
|
| 174 |
+
" OUTPUT_DIR = 'gs://{}/{}'.format(BUCKET, OUTPUT_DIR)\n",
|
| 175 |
+
" from google.colab import auth\n",
|
| 176 |
+
" auth.authenticate_user()\n",
|
| 177 |
+
"\n",
|
| 178 |
+
"if DO_DELETE:\n",
|
| 179 |
+
" try:\n",
|
| 180 |
+
" tf.gfile.DeleteRecursively(OUTPUT_DIR)\n",
|
| 181 |
+
" except:\n",
|
| 182 |
+
" # Doesn't matter if the directory didn't exist\n",
|
| 183 |
+
" pass\n",
|
| 184 |
+
"tf.gfile.MakeDirs(OUTPUT_DIR)\n",
|
| 185 |
+
"print('***** Model output directory: {} *****'.format(OUTPUT_DIR))\n"
|
| 186 |
+
],
|
| 187 |
+
"execution_count": 40,
|
| 188 |
+
"outputs": [
|
| 189 |
+
{
|
| 190 |
+
"output_type": "stream",
|
| 191 |
+
"text": [
|
| 192 |
+
"***** Model output directory: gs://bert-tfhub/aclImdb_v1 *****\n"
|
| 193 |
+
],
|
| 194 |
+
"name": "stdout"
|
| 195 |
+
}
|
| 196 |
+
]
|
| 197 |
+
},
|
| 198 |
+
{
|
| 199 |
+
"metadata": {
|
| 200 |
+
"id": "pmFYvkylMwXn",
|
| 201 |
+
"colab_type": "text"
|
| 202 |
+
},
|
| 203 |
+
"cell_type": "markdown",
|
| 204 |
+
"source": [
|
| 205 |
+
"#Data"
|
| 206 |
+
]
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"metadata": {
|
| 210 |
+
"id": "MC_w8SRqN0fr",
|
| 211 |
+
"colab_type": "text"
|
| 212 |
+
},
|
| 213 |
+
"cell_type": "markdown",
|
| 214 |
+
"source": [
|
| 215 |
+
"First, let's download the dataset, hosted by Stanford. The code below, which downloads, extracts, and imports the IMDB Large Movie Review Dataset, is borrowed from [this Tensorflow tutorial](https://www.tensorflow.org/hub/tutorials/text_classification_with_tf_hub)."
|
| 216 |
+
]
|
| 217 |
+
},
|
| 218 |
+
{
|
| 219 |
+
"metadata": {
|
| 220 |
+
"id": "fom_ff20gyy6",
|
| 221 |
+
"colab_type": "code",
|
| 222 |
+
"colab": {}
|
| 223 |
+
},
|
| 224 |
+
"cell_type": "code",
|
| 225 |
+
"source": [
|
| 226 |
+
"from tensorflow import keras\n",
|
| 227 |
+
"import os\n",
|
| 228 |
+
"import re\n",
|
| 229 |
+
"\n",
|
| 230 |
+
"# Load all files from a directory in a DataFrame.\n",
|
| 231 |
+
"def load_directory_data(directory):\n",
|
| 232 |
+
" data = {}\n",
|
| 233 |
+
" data[\"sentence\"] = []\n",
|
| 234 |
+
" data[\"sentiment\"] = []\n",
|
| 235 |
+
" for file_path in os.listdir(directory):\n",
|
| 236 |
+
" with tf.gfile.GFile(os.path.join(directory, file_path), \"r\") as f:\n",
|
| 237 |
+
" data[\"sentence\"].append(f.read())\n",
|
| 238 |
+
" data[\"sentiment\"].append(re.match(\"\\d+_(\\d+)\\.txt\", file_path).group(1))\n",
|
| 239 |
+
" return pd.DataFrame.from_dict(data)\n",
|
| 240 |
+
"\n",
|
| 241 |
+
"# Merge positive and negative examples, add a polarity column and shuffle.\n",
|
| 242 |
+
"def load_dataset(directory):\n",
|
| 243 |
+
" pos_df = load_directory_data(os.path.join(directory, \"pos\"))\n",
|
| 244 |
+
" neg_df = load_directory_data(os.path.join(directory, \"neg\"))\n",
|
| 245 |
+
" pos_df[\"polarity\"] = 1\n",
|
| 246 |
+
" neg_df[\"polarity\"] = 0\n",
|
| 247 |
+
" return pd.concat([pos_df, neg_df]).sample(frac=1).reset_index(drop=True)\n",
|
| 248 |
+
"\n",
|
| 249 |
+
"# Download and process the dataset files.\n",
|
| 250 |
+
"def download_and_load_datasets(force_download=False):\n",
|
| 251 |
+
" dataset = tf.keras.utils.get_file(\n",
|
| 252 |
+
" fname=\"aclImdb.tar.gz\", \n",
|
| 253 |
+
" origin=\"http://ai.stanford.edu/~amaas/data/sentiment/aclImdb_v1.tar.gz\", \n",
|
| 254 |
+
" extract=True)\n",
|
| 255 |
+
" \n",
|
| 256 |
+
" train_df = load_dataset(os.path.join(os.path.dirname(dataset), \n",
|
| 257 |
+
" \"aclImdb\", \"train\"))\n",
|
| 258 |
+
" test_df = load_dataset(os.path.join(os.path.dirname(dataset), \n",
|
| 259 |
+
" \"aclImdb\", \"test\"))\n",
|
| 260 |
+
" \n",
|
| 261 |
+
" return train_df, test_df\n"
|
| 262 |
+
],
|
| 263 |
+
"execution_count": 0,
|
| 264 |
+
"outputs": []
|
| 265 |
+
},
|
| 266 |
+
{
|
| 267 |
+
"metadata": {
|
| 268 |
+
"id": "2abfwdn-g135",
|
| 269 |
+
"colab_type": "code",
|
| 270 |
+
"colab": {}
|
| 271 |
+
},
|
| 272 |
+
"cell_type": "code",
|
| 273 |
+
"source": [
|
| 274 |
+
"train, test = download_and_load_datasets()"
|
| 275 |
+
],
|
| 276 |
+
"execution_count": 0,
|
| 277 |
+
"outputs": []
|
| 278 |
+
},
|
| 279 |
+
{
|
| 280 |
+
"metadata": {
|
| 281 |
+
"id": "XA8WHJgzhIZf",
|
| 282 |
+
"colab_type": "text"
|
| 283 |
+
},
|
| 284 |
+
"cell_type": "markdown",
|
| 285 |
+
"source": [
|
| 286 |
+
"To keep training fast, we'll take a sample of 5000 train and test examples, respectively."
|
| 287 |
+
]
|
| 288 |
+
},
|
| 289 |
+
{
|
| 290 |
+
"metadata": {
|
| 291 |
+
"id": "lw_F488eixTV",
|
| 292 |
+
"colab_type": "code",
|
| 293 |
+
"colab": {}
|
| 294 |
+
},
|
| 295 |
+
"cell_type": "code",
|
| 296 |
+
"source": [
|
| 297 |
+
"train = train.sample(5000)\n",
|
| 298 |
+
"test = test.sample(5000)"
|
| 299 |
+
],
|
| 300 |
+
"execution_count": 0,
|
| 301 |
+
"outputs": []
|
| 302 |
+
},
|
| 303 |
+
{
|
| 304 |
+
"metadata": {
|
| 305 |
+
"id": "prRQM8pDi8xI",
|
| 306 |
+
"colab_type": "code",
|
| 307 |
+
"outputId": "34445cb8-2be0-4379-fdbc-7794091f6049",
|
| 308 |
+
"colab": {
|
| 309 |
+
"base_uri": "https://localhost:8080/",
|
| 310 |
+
"height": 34
|
| 311 |
+
}
|
| 312 |
+
},
|
| 313 |
+
"cell_type": "code",
|
| 314 |
+
"source": [
|
| 315 |
+
"train.columns"
|
| 316 |
+
],
|
| 317 |
+
"execution_count": 44,
|
| 318 |
+
"outputs": [
|
| 319 |
+
{
|
| 320 |
+
"output_type": "execute_result",
|
| 321 |
+
"data": {
|
| 322 |
+
"text/plain": [
|
| 323 |
+
"Index(['sentence', 'sentiment', 'polarity'], dtype='object')"
|
| 324 |
+
]
|
| 325 |
+
},
|
| 326 |
+
"metadata": {
|
| 327 |
+
"tags": []
|
| 328 |
+
},
|
| 329 |
+
"execution_count": 44
|
| 330 |
+
}
|
| 331 |
+
]
|
| 332 |
+
},
|
| 333 |
+
{
|
| 334 |
+
"metadata": {
|
| 335 |
+
"id": "sfRnHSz3iSXz",
|
| 336 |
+
"colab_type": "text"
|
| 337 |
+
},
|
| 338 |
+
"cell_type": "markdown",
|
| 339 |
+
"source": [
|
| 340 |
+
"For us, our input data is the 'sentence' column and our label is the 'polarity' column (0, 1 for negative and positive, respecitvely)"
|
| 341 |
+
]
|
| 342 |
+
},
|
| 343 |
+
{
|
| 344 |
+
"metadata": {
|
| 345 |
+
"id": "IuMOGwFui4it",
|
| 346 |
+
"colab_type": "code",
|
| 347 |
+
"colab": {}
|
| 348 |
+
},
|
| 349 |
+
"cell_type": "code",
|
| 350 |
+
"source": [
|
| 351 |
+
"DATA_COLUMN = 'sentence'\n",
|
| 352 |
+
"LABEL_COLUMN = 'polarity'\n",
|
| 353 |
+
"# label_list is the list of labels, i.e. True, False or 0, 1 or 'dog', 'cat'\n",
|
| 354 |
+
"label_list = [0, 1]"
|
| 355 |
+
],
|
| 356 |
+
"execution_count": 0,
|
| 357 |
+
"outputs": []
|
| 358 |
+
},
|
| 359 |
+
{
|
| 360 |
+
"metadata": {
|
| 361 |
+
"id": "V399W0rqNJ-Z",
|
| 362 |
+
"colab_type": "text"
|
| 363 |
+
},
|
| 364 |
+
"cell_type": "markdown",
|
| 365 |
+
"source": [
|
| 366 |
+
"#Data Preprocessing\n",
|
| 367 |
+
"We'll need to transform our data into a format BERT understands. This involves two steps. First, we create `InputExample`'s using the constructor provided in the BERT library.\n",
|
| 368 |
+
"\n",
|
| 369 |
+
"- `text_a` is the text we want to classify, which in this case, is the `Request` field in our Dataframe. \n",
|
| 370 |
+
"- `text_b` is used if we're training a model to understand the relationship between sentences (i.e. is `text_b` a translation of `text_a`? Is `text_b` an answer to the question asked by `text_a`?). This doesn't apply to our task, so we can leave `text_b` blank.\n",
|
| 371 |
+
"- `label` is the label for our example, i.e. True, False"
|
| 372 |
+
]
|
| 373 |
+
},
|
| 374 |
+
{
|
| 375 |
+
"metadata": {
|
| 376 |
+
"id": "p9gEt5SmM6i6",
|
| 377 |
+
"colab_type": "code",
|
| 378 |
+
"colab": {}
|
| 379 |
+
},
|
| 380 |
+
"cell_type": "code",
|
| 381 |
+
"source": [
|
| 382 |
+
"# Use the InputExample class from BERT's run_classifier code to create examples from the data\n",
|
| 383 |
+
"train_InputExamples = train.apply(lambda x: bert.run_classifier.InputExample(guid=None, # Globally unique ID for bookkeeping, unused in this example\n",
|
| 384 |
+
" text_a = x[DATA_COLUMN], \n",
|
| 385 |
+
" text_b = None, \n",
|
| 386 |
+
" label = x[LABEL_COLUMN]), axis = 1)\n",
|
| 387 |
+
"\n",
|
| 388 |
+
"test_InputExamples = test.apply(lambda x: bert.run_classifier.InputExample(guid=None, \n",
|
| 389 |
+
" text_a = x[DATA_COLUMN], \n",
|
| 390 |
+
" text_b = None, \n",
|
| 391 |
+
" label = x[LABEL_COLUMN]), axis = 1)"
|
| 392 |
+
],
|
| 393 |
+
"execution_count": 0,
|
| 394 |
+
"outputs": []
|
| 395 |
+
},
|
| 396 |
+
{
|
| 397 |
+
"metadata": {
|
| 398 |
+
"id": "SCZWZtKxObjh",
|
| 399 |
+
"colab_type": "text"
|
| 400 |
+
},
|
| 401 |
+
"cell_type": "markdown",
|
| 402 |
+
"source": [
|
| 403 |
+
"Next, we need to preprocess our data so that it matches the data BERT was trained on. For this, we'll need to do a couple of things (but don't worry--this is also included in the Python library):\n",
|
| 404 |
+
"\n",
|
| 405 |
+
"\n",
|
| 406 |
+
"1. Lowercase our text (if we're using a BERT lowercase model)\n",
|
| 407 |
+
"2. Tokenize it (i.e. \"sally says hi\" -> [\"sally\", \"says\", \"hi\"])\n",
|
| 408 |
+
"3. Break words into WordPieces (i.e. \"calling\" -> [\"call\", \"##ing\"])\n",
|
| 409 |
+
"4. Map our words to indexes using a vocab file that BERT provides\n",
|
| 410 |
+
"5. Add special \"CLS\" and \"SEP\" tokens (see the [readme](https://github.com/google-research/bert))\n",
|
| 411 |
+
"6. Append \"index\" and \"segment\" tokens to each input (see the [BERT paper](https://arxiv.org/pdf/1810.04805.pdf))\n",
|
| 412 |
+
"\n",
|
| 413 |
+
"Happily, we don't have to worry about most of these details.\n",
|
| 414 |
+
"\n",
|
| 415 |
+
"\n"
|
| 416 |
+
]
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"metadata": {
|
| 420 |
+
"id": "qMWiDtpyQSoU",
|
| 421 |
+
"colab_type": "text"
|
| 422 |
+
},
|
| 423 |
+
"cell_type": "markdown",
|
| 424 |
+
"source": [
|
| 425 |
+
"To start, we'll need to load a vocabulary file and lowercasing information directly from the BERT tf hub module:"
|
| 426 |
+
]
|
| 427 |
+
},
|
| 428 |
+
{
|
| 429 |
+
"metadata": {
|
| 430 |
+
"id": "IhJSe0QHNG7U",
|
| 431 |
+
"colab_type": "code",
|
| 432 |
+
"outputId": "20b28cc7-3cb3-4ce6-bfff-a7847ce3bbaa",
|
| 433 |
+
"colab": {
|
| 434 |
+
"base_uri": "https://localhost:8080/",
|
| 435 |
+
"height": 34
|
| 436 |
+
}
|
| 437 |
+
},
|
| 438 |
+
"cell_type": "code",
|
| 439 |
+
"source": [
|
| 440 |
+
"# This is a path to an uncased (all lowercase) version of BERT\n",
|
| 441 |
+
"BERT_MODEL_HUB = \"https://tfhub.dev/google/bert_uncased_L-12_H-768_A-12/1\"\n",
|
| 442 |
+
"\n",
|
| 443 |
+
"def create_tokenizer_from_hub_module():\n",
|
| 444 |
+
" \"\"\"Get the vocab file and casing info from the Hub module.\"\"\"\n",
|
| 445 |
+
" with tf.Graph().as_default():\n",
|
| 446 |
+
" bert_module = hub.Module(BERT_MODEL_HUB)\n",
|
| 447 |
+
" tokenization_info = bert_module(signature=\"tokenization_info\", as_dict=True)\n",
|
| 448 |
+
" with tf.Session() as sess:\n",
|
| 449 |
+
" vocab_file, do_lower_case = sess.run([tokenization_info[\"vocab_file\"],\n",
|
| 450 |
+
" tokenization_info[\"do_lower_case\"]])\n",
|
| 451 |
+
" \n",
|
| 452 |
+
" return bert.tokenization.FullTokenizer(\n",
|
| 453 |
+
" vocab_file=vocab_file, do_lower_case=do_lower_case)\n",
|
| 454 |
+
"\n",
|
| 455 |
+
"tokenizer = create_tokenizer_from_hub_module()"
|
| 456 |
+
],
|
| 457 |
+
"execution_count": 47,
|
| 458 |
+
"outputs": [
|
| 459 |
+
{
|
| 460 |
+
"output_type": "stream",
|
| 461 |
+
"text": [
|
| 462 |
+
"INFO:tensorflow:Saver not created because there are no variables in the graph to restore\n"
|
| 463 |
+
],
|
| 464 |
+
"name": "stdout"
|
| 465 |
+
}
|
| 466 |
+
]
|
| 467 |
+
},
|
| 468 |
+
{
|
| 469 |
+
"metadata": {
|
| 470 |
+
"id": "z4oFkhpZBDKm",
|
| 471 |
+
"colab_type": "text"
|
| 472 |
+
},
|
| 473 |
+
"cell_type": "markdown",
|
| 474 |
+
"source": [
|
| 475 |
+
"Great--we just learned that the BERT model we're using expects lowercase data (that's what stored in tokenization_info[\"do_lower_case\"]) and we also loaded BERT's vocab file. We also created a tokenizer, which breaks words into word pieces:"
|
| 476 |
+
]
|
| 477 |
+
},
|
| 478 |
+
{
|
| 479 |
+
"metadata": {
|
| 480 |
+
"id": "dsBo6RCtQmwx",
|
| 481 |
+
"colab_type": "code",
|
| 482 |
+
"outputId": "9af8c917-90ec-4fe9-897b-79dc89ca88e1",
|
| 483 |
+
"colab": {
|
| 484 |
+
"base_uri": "https://localhost:8080/",
|
| 485 |
+
"height": 221
|
| 486 |
+
}
|
| 487 |
+
},
|
| 488 |
+
"cell_type": "code",
|
| 489 |
+
"source": [
|
| 490 |
+
"tokenizer.tokenize(\"This here's an example of using the BERT tokenizer\")"
|
| 491 |
+
],
|
| 492 |
+
"execution_count": 48,
|
| 493 |
+
"outputs": [
|
| 494 |
+
{
|
| 495 |
+
"output_type": "execute_result",
|
| 496 |
+
"data": {
|
| 497 |
+
"text/plain": [
|
| 498 |
+
"['this',\n",
|
| 499 |
+
" 'here',\n",
|
| 500 |
+
" \"'\",\n",
|
| 501 |
+
" 's',\n",
|
| 502 |
+
" 'an',\n",
|
| 503 |
+
" 'example',\n",
|
| 504 |
+
" 'of',\n",
|
| 505 |
+
" 'using',\n",
|
| 506 |
+
" 'the',\n",
|
| 507 |
+
" 'bert',\n",
|
| 508 |
+
" 'token',\n",
|
| 509 |
+
" '##izer']"
|
| 510 |
+
]
|
| 511 |
+
},
|
| 512 |
+
"metadata": {
|
| 513 |
+
"tags": []
|
| 514 |
+
},
|
| 515 |
+
"execution_count": 48
|
| 516 |
+
}
|
| 517 |
+
]
|
| 518 |
+
},
|
| 519 |
+
{
|
| 520 |
+
"metadata": {
|
| 521 |
+
"id": "0OEzfFIt6GIc",
|
| 522 |
+
"colab_type": "text"
|
| 523 |
+
},
|
| 524 |
+
"cell_type": "markdown",
|
| 525 |
+
"source": [
|
| 526 |
+
"Using our tokenizer, we'll call `run_classifier.convert_examples_to_features` on our InputExamples to convert them into features BERT understands."
|
| 527 |
+
]
|
| 528 |
+
},
|
| 529 |
+
{
|
| 530 |
+
"metadata": {
|
| 531 |
+
"id": "LL5W8gEGRTAf",
|
| 532 |
+
"colab_type": "code",
|
| 533 |
+
"outputId": "65001dda-155b-48fc-b5fc-1e4cabc8dfbf",
|
| 534 |
+
"colab": {
|
| 535 |
+
"base_uri": "https://localhost:8080/",
|
| 536 |
+
"height": 1261
|
| 537 |
+
}
|
| 538 |
+
},
|
| 539 |
+
"cell_type": "code",
|
| 540 |
+
"source": [
|
| 541 |
+
"# We'll set sequences to be at most 128 tokens long.\n",
|
| 542 |
+
"MAX_SEQ_LENGTH = 128\n",
|
| 543 |
+
"# Convert our train and test features to InputFeatures that BERT understands.\n",
|
| 544 |
+
"train_features = bert.run_classifier.convert_examples_to_features(train_InputExamples, label_list, MAX_SEQ_LENGTH, tokenizer)\n",
|
| 545 |
+
"test_features = bert.run_classifier.convert_examples_to_features(test_InputExamples, label_list, MAX_SEQ_LENGTH, tokenizer)"
|
| 546 |
+
],
|
| 547 |
+
"execution_count": 49,
|
| 548 |
+
"outputs": [
|
| 549 |
+
{
|
| 550 |
+
"output_type": "stream",
|
| 551 |
+
"text": [
|
| 552 |
+
"INFO:tensorflow:Writing example 0 of 5000\n",
|
| 553 |
+
"INFO:tensorflow:*** Example ***\n",
|
| 554 |
+
"INFO:tensorflow:guid: None\n",
|
| 555 |
+
"INFO:tensorflow:tokens: [CLS] i ' m watching this on the sci - fi channel right now . it ' s so horrible i can ' t stop watching it ! i ' m a video ##grapher and this movie makes me sad . i feel bad for anyone associated with this movie . some of the camera work is good . most is very questionable . there are a few decent actors in the flick . too bad they ' re surrounded by what must have been the director ' s relatives . that ' s the only way they could have been qualified to be in a movie ! music was a little better than the acting . if you get around to watching this i hope it [SEP]\n",
|
| 556 |
+
"INFO:tensorflow:input_ids: 101 1045 1005 1049 3666 2023 2006 1996 16596 1011 10882 3149 2157 2085 1012 2009 1005 1055 2061 9202 1045 2064 1005 1056 2644 3666 2009 999 1045 1005 1049 1037 2678 18657 1998 2023 3185 3084 2033 6517 1012 1045 2514 2919 2005 3087 3378 2007 2023 3185 1012 2070 1997 1996 4950 2147 2003 2204 1012 2087 2003 2200 21068 1012 2045 2024 1037 2261 11519 5889 1999 1996 17312 1012 2205 2919 2027 1005 2128 5129 2011 2054 2442 2031 2042 1996 2472 1005 1055 9064 1012 2008 1005 1055 1996 2069 2126 2027 2071 2031 2042 4591 2000 2022 1999 1037 3185 999 2189 2001 1037 2210 2488 2084 1996 3772 1012 2065 2017 2131 2105 2000 3666 2023 1045 3246 2009 102\n",
|
| 557 |
+
"INFO:tensorflow:input_mask: 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1\n",
|
| 558 |
+
"INFO:tensorflow:segment_ids: 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n",
|
| 559 |
+
"INFO:tensorflow:label: 0 (id = 0)\n",
|
| 560 |
+
"INFO:tensorflow:*** Example ***\n",
|
| 561 |
+
"INFO:tensorflow:guid: None\n",
|
| 562 |
+
"INFO:tensorflow:tokens: [CLS] i have been a fan of pushing dai ##sies since the very beginning . it is wonderful ##ly thought up , and bryan fuller has the most remarkable ideas for this show . < br / > < br / > it is unbelievable on how much tv has been needing a creative , original show like pushing dai ##sies . it is a huge relief to see a show , that is unlike the rest , where as , if you compared it to some of the newer shows , such as scrub ##s and house , you would see the similarities , and it does get ted ##ious at moments to see shows so close in identity . < br / > < br [SEP]\n",
|
| 563 |
+
"INFO:tensorflow:input_ids: 101 1045 2031 2042 1037 5470 1997 6183 18765 14625 2144 1996 2200 2927 1012 2009 2003 6919 2135 2245 2039 1010 1998 8527 12548 2038 1996 2087 9487 4784 2005 2023 2265 1012 1026 7987 1013 1028 1026 7987 1013 1028 2009 2003 23653 2006 2129 2172 2694 2038 2042 11303 1037 5541 1010 2434 2265 2066 6183 18765 14625 1012 2009 2003 1037 4121 4335 2000 2156 1037 2265 1010 2008 2003 4406 1996 2717 1010 2073 2004 1010 2065 2017 4102 2009 2000 2070 1997 1996 10947 3065 1010 2107 2004 18157 2015 1998 2160 1010 2017 2052 2156 1996 12319 1010 1998 2009 2515 2131 6945 6313 2012 5312 2000 2156 3065 2061 2485 1999 4767 1012 1026 7987 1013 1028 1026 7987 102\n",
|
| 564 |
+
"INFO:tensorflow:input_mask: 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1\n",
|
| 565 |
+
"INFO:tensorflow:segment_ids: 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n",
|
| 566 |
+
"INFO:tensorflow:label: 1 (id = 1)\n",
|
| 567 |
+
"INFO:tensorflow:*** Example ***\n",
|
| 568 |
+
"INFO:tensorflow:guid: None\n",
|
| 569 |
+
"INFO:tensorflow:tokens: [CLS] this movie starts out promising ##ly , with an early scene in which frank morgan advises against gary cooper ' s marriage to his daughter , anita louise . frank morgan , playing an una ##bas ##hed gold - digger , loudly complain ##s to cooper about his perceived pen ##ury at the hands of his family - including his daughter , anita louise . i am a fan of all 3 actors . frank morgan is ( to my mind ) a hollywood treasure , cooper a legend , and louise a very lovely , versatile and under - appreciated actress seldom seen in the leading role . i also have nothing against teresa wright , and while not blessed with great range , she [SEP]\n",
|
| 570 |
+
"INFO:tensorflow:input_ids: 101 2023 3185 4627 2041 10015 2135 1010 2007 2019 2220 3496 1999 2029 3581 5253 25453 2114 5639 6201 1005 1055 3510 2000 2010 2684 1010 12918 8227 1012 3581 5253 1010 2652 2019 14477 22083 9072 2751 1011 28661 1010 9928 17612 2015 2000 6201 2055 2010 8690 7279 13098 2012 1996 2398 1997 2010 2155 1011 2164 2010 2684 1010 12918 8227 1012 1045 2572 1037 5470 1997 2035 1017 5889 1012 3581 5253 2003 1006 2000 2026 2568 1007 1037 5365 8813 1010 6201 1037 5722 1010 1998 8227 1037 2200 8403 1010 22979 1998 2104 1011 12315 3883 15839 2464 1999 1996 2877 2535 1012 1045 2036 2031 2498 2114 12409 6119 1010 1998 2096 2025 10190 2007 2307 2846 1010 2016 102\n",
|
| 571 |
+
"INFO:tensorflow:input_mask: 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1\n",
|
| 572 |
+
"INFO:tensorflow:segment_ids: 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n",
|
| 573 |
+
"INFO:tensorflow:label: 0 (id = 0)\n",
|
| 574 |
+
"INFO:tensorflow:*** Example ***\n",
|
| 575 |
+
"INFO:tensorflow:guid: None\n",
|
| 576 |
+
"INFO:tensorflow:tokens: [CLS] i was over ##taken by the emotion . un ##for ##get ##table rendering of a wartime story which is unknown to most people . the performances were fault ##less and outstanding . [SEP]\n",
|
| 577 |
+
"INFO:tensorflow:input_ids: 101 1045 2001 2058 25310 2011 1996 7603 1012 4895 29278 18150 10880 14259 1997 1037 12498 2466 2029 2003 4242 2000 2087 2111 1012 1996 4616 2020 6346 3238 1998 5151 1012 102 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n",
|
| 578 |
+
"INFO:tensorflow:input_mask: 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n",
|
| 579 |
+
"INFO:tensorflow:segment_ids: 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n",
|
| 580 |
+
"INFO:tensorflow:label: 1 (id = 1)\n",
|
| 581 |
+
"INFO:tensorflow:*** Example ***\n",
|
| 582 |
+
"INFO:tensorflow:guid: None\n",
|
| 583 |
+
"INFO:tensorflow:tokens: [CLS] soldier blue is a movie with pre ##tension ##s : pre ##tension ##s to be some sort of profound statement on man ' s inhuman ##ity to man , on the white man ' s exploitation of and brutality towards indigenous peoples ; a biting , un ##fl ##in ##ching and sar ##don ##ic commentary on the horrors of vietnam . well , sorry , but it fails mis ##era ##bly to be any of those things . what soldier blue actually is is per ##nic ##ious , tri ##te , badly made , dish ##ones ##t rubbish . < br / > < br / > another reviewer here hit the nail on the head in saying that it appears to be a hybrid of [SEP]\n",
|
| 584 |
+
"INFO:tensorflow:input_ids: 101 5268 2630 2003 1037 3185 2007 3653 29048 2015 1024 3653 29048 2015 2000 2022 2070 4066 1997 13769 4861 2006 2158 1005 1055 29582 3012 2000 2158 1010 2006 1996 2317 2158 1005 1055 14427 1997 1998 24083 2875 6284 7243 1025 1037 12344 1010 4895 10258 2378 8450 1998 18906 5280 2594 8570 2006 1996 22812 1997 5148 1012 2092 1010 3374 1010 2021 2009 11896 28616 6906 6321 2000 2022 2151 1997 2216 2477 1012 2054 5268 2630 2941 2003 2003 2566 8713 6313 1010 13012 2618 1010 6649 2081 1010 9841 21821 2102 29132 1012 1026 7987 1013 1028 1026 7987 1013 1028 2178 12027 2182 2718 1996 13774 2006 1996 2132 1999 3038 2008 2009 3544 2000 2022 1037 8893 1997 102\n",
|
| 585 |
+
"INFO:tensorflow:input_mask: 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1\n",
|
| 586 |
+
"INFO:tensorflow:segment_ids: 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n",
|
| 587 |
+
"INFO:tensorflow:label: 0 (id = 0)\n",
|
| 588 |
+
"INFO:tensorflow:Writing example 0 of 5000\n",
|
| 589 |
+
"INFO:tensorflow:*** Example ***\n",
|
| 590 |
+
"INFO:tensorflow:guid: None\n",
|
| 591 |
+
"INFO:tensorflow:tokens: [CLS] i just watched this today on tv . it was on abc ' s sunday afternoon movie . < br / > < br / > this wasn ' t a very good movie , but for a low budget independent film like this , it was okay . there is some suspense in it , but there are so many bad qualities that really bring the movie down . the script is pretty lame , and the plot elements aren ' t very realistic , such as the way a 911 operator would laugh and hang up when someone is reporting a murder . i don ' t know what the writer was thinking when they came up with that idea , but it isn [SEP]\n",
|
| 592 |
+
"INFO:tensorflow:input_ids: 101 1045 2074 3427 2023 2651 2006 2694 1012 2009 2001 2006 5925 1005 1055 4465 5027 3185 1012 1026 7987 1013 1028 1026 7987 1013 1028 2023 2347 1005 1056 1037 2200 2204 3185 1010 2021 2005 1037 2659 5166 2981 2143 2066 2023 1010 2009 2001 3100 1012 2045 2003 2070 23873 1999 2009 1010 2021 2045 2024 2061 2116 2919 11647 2008 2428 3288 1996 3185 2091 1012 1996 5896 2003 3492 20342 1010 1998 1996 5436 3787 4995 1005 1056 2200 12689 1010 2107 2004 1996 2126 1037 19989 6872 2052 4756 1998 6865 2039 2043 2619 2003 7316 1037 4028 1012 1045 2123 1005 1056 2113 2054 1996 3213 2001 3241 2043 2027 2234 2039 2007 2008 2801 1010 2021 2009 3475 102\n",
|
| 593 |
+
"INFO:tensorflow:input_mask: 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1\n",
|
| 594 |
+
"INFO:tensorflow:segment_ids: 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n",
|
| 595 |
+
"INFO:tensorflow:label: 0 (id = 0)\n",
|
| 596 |
+
"INFO:tensorflow:*** Example ***\n",
|
| 597 |
+
"INFO:tensorflow:guid: None\n",
|
| 598 |
+
"INFO:tensorflow:tokens: [CLS] from hardly alien sounding lasers , to an elementary school style shuttle crash , \" night ##be ##ast \" is better classified as a far ##cic ##al mix of fake blood and bare chest . the almost pornographic style of the film seems to be a failed attempt to recover from a lack of co ##hesive or effective story . the acting however is not nearly as beast ##ly , many of the young , aspiring , actors ad ##mir ##ably showcase a hidden talent . particularly don lei ##fer ##t and jamie ze ##mare ##l , who shed a well needed sha ##rd of light on this otherwise terrible film . night ##be ##ast would have never shown up on set had he known the [SEP]\n",
|
| 599 |
+
"INFO:tensorflow:input_ids: 101 2013 6684 7344 9391 23965 1010 2000 2019 4732 2082 2806 10382 5823 1010 1000 2305 4783 14083 1000 2003 2488 6219 2004 1037 2521 19053 2389 4666 1997 8275 2668 1998 6436 3108 1012 1996 2471 26932 2806 1997 1996 2143 3849 2000 2022 1037 3478 3535 2000 8980 2013 1037 3768 1997 2522 21579 2030 4621 2466 1012 1996 3772 2174 2003 2025 3053 2004 6841 2135 1010 2116 1997 1996 2402 1010 22344 1010 5889 4748 14503 8231 13398 1037 5023 5848 1012 3391 2123 26947 7512 2102 1998 6175 27838 24376 2140 1010 2040 8328 1037 2092 2734 21146 4103 1997 2422 2006 2023 4728 6659 2143 1012 2305 4783 14083 2052 2031 2196 3491 2039 2006 2275 2018 2002 2124 1996 102\n",
|
| 600 |
+
"INFO:tensorflow:input_mask: 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1\n",
|
| 601 |
+
"INFO:tensorflow:segment_ids: 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n",
|
| 602 |
+
"INFO:tensorflow:label: 0 (id = 0)\n",
|
| 603 |
+
"INFO:tensorflow:*** Example ***\n",
|
| 604 |
+
"INFO:tensorflow:guid: None\n",
|
| 605 |
+
"INFO:tensorflow:tokens: [CLS] here we have the in ##imi ##table charlie chaplin for ##sa ##king his slap ##stick past to tackle the serious subject of anti - semi ##tism , and into ##ler ##ance in general . he portrays two characters - the sweet , innocent jewish barber - a war veteran , and the ravi ##ng and ruthless dictator , aden ##oid h ##yn ##kel . the jewish ghetto in this country is not safe for long , due to the w ##him ##s of h ##yn ##kel and his armed thugs , who routinely rough up its residents , or leave them alone , dependent upon his mood that day or week . the barber is among them , but is befriended by his former commanding officer [SEP]\n",
|
| 606 |
+
"INFO:tensorflow:input_ids: 101 2182 2057 2031 1996 1999 27605 10880 4918 23331 2005 3736 6834 2010 14308 21354 2627 2000 11147 1996 3809 3395 1997 3424 1011 4100 17456 1010 1998 2046 3917 6651 1999 2236 1012 2002 17509 2048 3494 1011 1996 4086 1010 7036 3644 13362 1011 1037 2162 8003 1010 1998 1996 16806 3070 1998 18101 21237 1010 16298 9314 1044 6038 11705 1012 1996 3644 17276 1999 2023 2406 2003 2025 3647 2005 2146 1010 2349 2000 1996 1059 14341 2015 1997 1044 6038 11705 1998 2010 4273 24106 1010 2040 19974 5931 2039 2049 3901 1010 2030 2681 2068 2894 1010 7790 2588 2010 6888 2008 2154 2030 2733 1012 1996 13362 2003 2426 2068 1010 2021 2003 23386 2011 2010 2280 7991 2961 102\n",
|
| 607 |
+
"INFO:tensorflow:input_mask: 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1\n",
|
| 608 |
+
"INFO:tensorflow:segment_ids: 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n",
|
| 609 |
+
"INFO:tensorflow:label: 1 (id = 1)\n",
|
| 610 |
+
"INFO:tensorflow:*** Example ***\n",
|
| 611 |
+
"INFO:tensorflow:guid: None\n",
|
| 612 |
+
"INFO:tensorflow:tokens: [CLS] i really hated this movie and it ' s the first movie written by stephen king that i didn ' t finish . i was truly disappointed , it was the worst crap i ' ve ever seen . what were you thinking making three hours out of it ? it may have a quite good story , but actors ? no . suspense ? no . romance ? no . horror ? no . it didn ' t have anything . < br / > < br / > it ' s got this strange , crazy science man with einstein - hair , the classic thing . not real at all . and a man keep getting younger all the time . it seems [SEP]\n",
|
| 613 |
+
"INFO:tensorflow:input_ids: 101 1045 2428 6283 2023 3185 1998 2009 1005 1055 1996 2034 3185 2517 2011 4459 2332 2008 1045 2134 1005 1056 3926 1012 1045 2001 5621 9364 1010 2009 2001 1996 5409 10231 1045 1005 2310 2412 2464 1012 2054 2020 2017 3241 2437 2093 2847 2041 1997 2009 1029 2009 2089 2031 1037 3243 2204 2466 1010 2021 5889 1029 2053 1012 23873 1029 2053 1012 7472 1029 2053 1012 5469 1029 2053 1012 2009 2134 1005 1056 2031 2505 1012 1026 7987 1013 1028 1026 7987 1013 1028 2009 1005 1055 2288 2023 4326 1010 4689 2671 2158 2007 15313 1011 2606 1010 1996 4438 2518 1012 2025 2613 2012 2035 1012 1998 1037 2158 2562 2893 3920 2035 1996 2051 1012 2009 3849 102\n",
|
| 614 |
+
"INFO:tensorflow:input_mask: 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1\n",
|
| 615 |
+
"INFO:tensorflow:segment_ids: 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n",
|
| 616 |
+
"INFO:tensorflow:label: 0 (id = 0)\n",
|
| 617 |
+
"INFO:tensorflow:*** Example ***\n",
|
| 618 |
+
"INFO:tensorflow:guid: None\n",
|
| 619 |
+
"INFO:tensorflow:tokens: [CLS] story chinese tall story tells the story of righteous monk trip ##ita ##ka , who , along with his guardians monkey , sandy and pigs ##y make their journey west on a quest to recover ancient sutra ##s , finally , they reach the final leg of their journey in sha ##che city but all is not as it seems when the city is attacked by evil tree demons . monkey tries his best to battle them but is overwhelmed , knowing his master is in grave danger , he uses his trust ##y golden staff to thrust trip ##ita ##ka to safety . < br / > < br / > the monk ends up being knocked out when he land and when he wakes [SEP]\n",
|
| 620 |
+
"INFO:tensorflow:input_ids: 101 2466 2822 4206 2466 4136 1996 2466 1997 19556 8284 4440 6590 2912 1010 2040 1010 2247 2007 2010 14240 10608 1010 7525 1998 14695 2100 2191 2037 4990 2225 2006 1037 8795 2000 8980 3418 26567 2015 1010 2633 1010 2027 3362 1996 2345 4190 1997 2037 4990 1999 21146 5403 2103 2021 2035 2003 2025 2004 2009 3849 2043 1996 2103 2003 4457 2011 4763 3392 7942 1012 10608 5363 2010 2190 2000 2645 2068 2021 2003 13394 1010 4209 2010 3040 2003 1999 6542 5473 1010 2002 3594 2010 3404 2100 3585 3095 2000 7400 4440 6590 2912 2000 3808 1012 1026 7987 1013 1028 1026 7987 1013 1028 1996 8284 4515 2039 2108 6573 2041 2043 2002 2455 1998 2043 2002 17507 102\n",
|
| 621 |
+
"INFO:tensorflow:input_mask: 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1\n",
|
| 622 |
+
"INFO:tensorflow:segment_ids: 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n",
|
| 623 |
+
"INFO:tensorflow:label: 1 (id = 1)\n"
|
| 624 |
+
],
|
| 625 |
+
"name": "stdout"
|
| 626 |
+
}
|
| 627 |
+
]
|
| 628 |
+
},
|
| 629 |
+
{
|
| 630 |
+
"metadata": {
|
| 631 |
+
"id": "ccp5trMwRtmr",
|
| 632 |
+
"colab_type": "text"
|
| 633 |
+
},
|
| 634 |
+
"cell_type": "markdown",
|
| 635 |
+
"source": [
|
| 636 |
+
"#Creating a model\n",
|
| 637 |
+
"\n",
|
| 638 |
+
"Now that we've prepared our data, let's focus on building a model. `create_model` does just this below. First, it loads the BERT tf hub module again (this time to extract the computation graph). Next, it creates a single new layer that will be trained to adapt BERT to our sentiment task (i.e. classifying whether a movie review is positive or negative). This strategy of using a mostly trained model is called [fine-tuning](http://wiki.fast.ai/index.php/Fine_tuning)."
|
| 639 |
+
]
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"metadata": {
|
| 643 |
+
"id": "6o2a5ZIvRcJq",
|
| 644 |
+
"colab_type": "code",
|
| 645 |
+
"colab": {}
|
| 646 |
+
},
|
| 647 |
+
"cell_type": "code",
|
| 648 |
+
"source": [
|
| 649 |
+
"def create_model(is_predicting, input_ids, input_mask, segment_ids, labels,\n",
|
| 650 |
+
" num_labels):\n",
|
| 651 |
+
" \"\"\"Creates a classification model.\"\"\"\n",
|
| 652 |
+
"\n",
|
| 653 |
+
" bert_module = hub.Module(\n",
|
| 654 |
+
" BERT_MODEL_HUB,\n",
|
| 655 |
+
" trainable=True)\n",
|
| 656 |
+
" bert_inputs = dict(\n",
|
| 657 |
+
" input_ids=input_ids,\n",
|
| 658 |
+
" input_mask=input_mask,\n",
|
| 659 |
+
" segment_ids=segment_ids)\n",
|
| 660 |
+
" bert_outputs = bert_module(\n",
|
| 661 |
+
" inputs=bert_inputs,\n",
|
| 662 |
+
" signature=\"tokens\",\n",
|
| 663 |
+
" as_dict=True)\n",
|
| 664 |
+
"\n",
|
| 665 |
+
" # Use \"pooled_output\" for classification tasks on an entire sentence.\n",
|
| 666 |
+
" # Use \"sequence_outputs\" for token-level output.\n",
|
| 667 |
+
" output_layer = bert_outputs[\"pooled_output\"]\n",
|
| 668 |
+
"\n",
|
| 669 |
+
" hidden_size = output_layer.shape[-1].value\n",
|
| 670 |
+
"\n",
|
| 671 |
+
" # Create our own layer to tune for politeness data.\n",
|
| 672 |
+
" output_weights = tf.get_variable(\n",
|
| 673 |
+
" \"output_weights\", [num_labels, hidden_size],\n",
|
| 674 |
+
" initializer=tf.truncated_normal_initializer(stddev=0.02))\n",
|
| 675 |
+
"\n",
|
| 676 |
+
" output_bias = tf.get_variable(\n",
|
| 677 |
+
" \"output_bias\", [num_labels], initializer=tf.zeros_initializer())\n",
|
| 678 |
+
"\n",
|
| 679 |
+
" with tf.variable_scope(\"loss\"):\n",
|
| 680 |
+
"\n",
|
| 681 |
+
" # Dropout helps prevent overfitting\n",
|
| 682 |
+
" output_layer = tf.nn.dropout(output_layer, keep_prob=0.9)\n",
|
| 683 |
+
"\n",
|
| 684 |
+
" logits = tf.matmul(output_layer, output_weights, transpose_b=True)\n",
|
| 685 |
+
" logits = tf.nn.bias_add(logits, output_bias)\n",
|
| 686 |
+
" log_probs = tf.nn.log_softmax(logits, axis=-1)\n",
|
| 687 |
+
"\n",
|
| 688 |
+
" # Convert labels into one-hot encoding\n",
|
| 689 |
+
" one_hot_labels = tf.one_hot(labels, depth=num_labels, dtype=tf.float32)\n",
|
| 690 |
+
"\n",
|
| 691 |
+
" predicted_labels = tf.squeeze(tf.argmax(log_probs, axis=-1, output_type=tf.int32))\n",
|
| 692 |
+
" # If we're predicting, we want predicted labels and the probabiltiies.\n",
|
| 693 |
+
" if is_predicting:\n",
|
| 694 |
+
" return (predicted_labels, log_probs)\n",
|
| 695 |
+
"\n",
|
| 696 |
+
" # If we're train/eval, compute loss between predicted and actual label\n",
|
| 697 |
+
" per_example_loss = -tf.reduce_sum(one_hot_labels * log_probs, axis=-1)\n",
|
| 698 |
+
" loss = tf.reduce_mean(per_example_loss)\n",
|
| 699 |
+
" return (loss, predicted_labels, log_probs)\n"
|
| 700 |
+
],
|
| 701 |
+
"execution_count": 0,
|
| 702 |
+
"outputs": []
|
| 703 |
+
},
|
| 704 |
+
{
|
| 705 |
+
"metadata": {
|
| 706 |
+
"id": "qpE0ZIDOCQzE",
|
| 707 |
+
"colab_type": "text"
|
| 708 |
+
},
|
| 709 |
+
"cell_type": "markdown",
|
| 710 |
+
"source": [
|
| 711 |
+
"Next we'll wrap our model function in a `model_fn_builder` function that adapts our model to work for training, evaluation, and prediction."
|
| 712 |
+
]
|
| 713 |
+
},
|
| 714 |
+
{
|
| 715 |
+
"metadata": {
|
| 716 |
+
"id": "FnH-AnOQ9KKW",
|
| 717 |
+
"colab_type": "code",
|
| 718 |
+
"colab": {}
|
| 719 |
+
},
|
| 720 |
+
"cell_type": "code",
|
| 721 |
+
"source": [
|
| 722 |
+
"# model_fn_builder actually creates our model function\n",
|
| 723 |
+
"# using the passed parameters for num_labels, learning_rate, etc.\n",
|
| 724 |
+
"def model_fn_builder(num_labels, learning_rate, num_train_steps,\n",
|
| 725 |
+
" num_warmup_steps):\n",
|
| 726 |
+
" \"\"\"Returns `model_fn` closure for TPUEstimator.\"\"\"\n",
|
| 727 |
+
" def model_fn(features, labels, mode, params): # pylint: disable=unused-argument\n",
|
| 728 |
+
" \"\"\"The `model_fn` for TPUEstimator.\"\"\"\n",
|
| 729 |
+
"\n",
|
| 730 |
+
" input_ids = features[\"input_ids\"]\n",
|
| 731 |
+
" input_mask = features[\"input_mask\"]\n",
|
| 732 |
+
" segment_ids = features[\"segment_ids\"]\n",
|
| 733 |
+
" label_ids = features[\"label_ids\"]\n",
|
| 734 |
+
"\n",
|
| 735 |
+
" is_predicting = (mode == tf.estimator.ModeKeys.PREDICT)\n",
|
| 736 |
+
" \n",
|
| 737 |
+
" # TRAIN and EVAL\n",
|
| 738 |
+
" if not is_predicting:\n",
|
| 739 |
+
"\n",
|
| 740 |
+
" (loss, predicted_labels, log_probs) = create_model(\n",
|
| 741 |
+
" is_predicting, input_ids, input_mask, segment_ids, label_ids, num_labels)\n",
|
| 742 |
+
"\n",
|
| 743 |
+
" train_op = bert.optimization.create_optimizer(\n",
|
| 744 |
+
" loss, learning_rate, num_train_steps, num_warmup_steps, use_tpu=False)\n",
|
| 745 |
+
"\n",
|
| 746 |
+
" # Calculate evaluation metrics. \n",
|
| 747 |
+
" def metric_fn(label_ids, predicted_labels):\n",
|
| 748 |
+
" accuracy = tf.metrics.accuracy(label_ids, predicted_labels)\n",
|
| 749 |
+
" f1_score = tf.contrib.metrics.f1_score(\n",
|
| 750 |
+
" label_ids,\n",
|
| 751 |
+
" predicted_labels)\n",
|
| 752 |
+
" auc = tf.metrics.auc(\n",
|
| 753 |
+
" label_ids,\n",
|
| 754 |
+
" predicted_labels)\n",
|
| 755 |
+
" recall = tf.metrics.recall(\n",
|
| 756 |
+
" label_ids,\n",
|
| 757 |
+
" predicted_labels)\n",
|
| 758 |
+
" precision = tf.metrics.precision(\n",
|
| 759 |
+
" label_ids,\n",
|
| 760 |
+
" predicted_labels) \n",
|
| 761 |
+
" true_pos = tf.metrics.true_positives(\n",
|
| 762 |
+
" label_ids,\n",
|
| 763 |
+
" predicted_labels)\n",
|
| 764 |
+
" true_neg = tf.metrics.true_negatives(\n",
|
| 765 |
+
" label_ids,\n",
|
| 766 |
+
" predicted_labels) \n",
|
| 767 |
+
" false_pos = tf.metrics.false_positives(\n",
|
| 768 |
+
" label_ids,\n",
|
| 769 |
+
" predicted_labels) \n",
|
| 770 |
+
" false_neg = tf.metrics.false_negatives(\n",
|
| 771 |
+
" label_ids,\n",
|
| 772 |
+
" predicted_labels)\n",
|
| 773 |
+
" return {\n",
|
| 774 |
+
" \"eval_accuracy\": accuracy,\n",
|
| 775 |
+
" \"f1_score\": f1_score,\n",
|
| 776 |
+
" \"auc\": auc,\n",
|
| 777 |
+
" \"precision\": precision,\n",
|
| 778 |
+
" \"recall\": recall,\n",
|
| 779 |
+
" \"true_positives\": true_pos,\n",
|
| 780 |
+
" \"true_negatives\": true_neg,\n",
|
| 781 |
+
" \"false_positives\": false_pos,\n",
|
| 782 |
+
" \"false_negatives\": false_neg\n",
|
| 783 |
+
" }\n",
|
| 784 |
+
"\n",
|
| 785 |
+
" eval_metrics = metric_fn(label_ids, predicted_labels)\n",
|
| 786 |
+
"\n",
|
| 787 |
+
" if mode == tf.estimator.ModeKeys.TRAIN:\n",
|
| 788 |
+
" return tf.estimator.EstimatorSpec(mode=mode,\n",
|
| 789 |
+
" loss=loss,\n",
|
| 790 |
+
" train_op=train_op)\n",
|
| 791 |
+
" else:\n",
|
| 792 |
+
" return tf.estimator.EstimatorSpec(mode=mode,\n",
|
| 793 |
+
" loss=loss,\n",
|
| 794 |
+
" eval_metric_ops=eval_metrics)\n",
|
| 795 |
+
" else:\n",
|
| 796 |
+
" (predicted_labels, log_probs) = create_model(\n",
|
| 797 |
+
" is_predicting, input_ids, input_mask, segment_ids, label_ids, num_labels)\n",
|
| 798 |
+
"\n",
|
| 799 |
+
" predictions = {\n",
|
| 800 |
+
" 'probabilities': log_probs,\n",
|
| 801 |
+
" 'labels': predicted_labels\n",
|
| 802 |
+
" }\n",
|
| 803 |
+
" return tf.estimator.EstimatorSpec(mode, predictions=predictions)\n",
|
| 804 |
+
"\n",
|
| 805 |
+
" # Return the actual model function in the closure\n",
|
| 806 |
+
" return model_fn\n"
|
| 807 |
+
],
|
| 808 |
+
"execution_count": 0,
|
| 809 |
+
"outputs": []
|
| 810 |
+
},
|
| 811 |
+
{
|
| 812 |
+
"metadata": {
|
| 813 |
+
"id": "OjwJ4bTeWXD8",
|
| 814 |
+
"colab_type": "code",
|
| 815 |
+
"colab": {}
|
| 816 |
+
},
|
| 817 |
+
"cell_type": "code",
|
| 818 |
+
"source": [
|
| 819 |
+
"# Compute train and warmup steps from batch size\n",
|
| 820 |
+
"# These hyperparameters are copied from this colab notebook (https://colab.sandbox.google.com/github/tensorflow/tpu/blob/master/tools/colab/bert_finetuning_with_cloud_tpus.ipynb)\n",
|
| 821 |
+
"BATCH_SIZE = 32\n",
|
| 822 |
+
"LEARNING_RATE = 2e-5\n",
|
| 823 |
+
"NUM_TRAIN_EPOCHS = 3.0\n",
|
| 824 |
+
"# Warmup is a period of time where hte learning rate \n",
|
| 825 |
+
"# is small and gradually increases--usually helps training.\n",
|
| 826 |
+
"WARMUP_PROPORTION = 0.1\n",
|
| 827 |
+
"# Model configs\n",
|
| 828 |
+
"SAVE_CHECKPOINTS_STEPS = 500\n",
|
| 829 |
+
"SAVE_SUMMARY_STEPS = 100"
|
| 830 |
+
],
|
| 831 |
+
"execution_count": 0,
|
| 832 |
+
"outputs": []
|
| 833 |
+
},
|
| 834 |
+
{
|
| 835 |
+
"metadata": {
|
| 836 |
+
"id": "emHf9GhfWBZ_",
|
| 837 |
+
"colab_type": "code",
|
| 838 |
+
"colab": {}
|
| 839 |
+
},
|
| 840 |
+
"cell_type": "code",
|
| 841 |
+
"source": [
|
| 842 |
+
"# Compute # train and warmup steps from batch size\n",
|
| 843 |
+
"num_train_steps = int(len(train_features) / BATCH_SIZE * NUM_TRAIN_EPOCHS)\n",
|
| 844 |
+
"num_warmup_steps = int(num_train_steps * WARMUP_PROPORTION)"
|
| 845 |
+
],
|
| 846 |
+
"execution_count": 0,
|
| 847 |
+
"outputs": []
|
| 848 |
+
},
|
| 849 |
+
{
|
| 850 |
+
"metadata": {
|
| 851 |
+
"id": "oEJldMr3WYZa",
|
| 852 |
+
"colab_type": "code",
|
| 853 |
+
"colab": {}
|
| 854 |
+
},
|
| 855 |
+
"cell_type": "code",
|
| 856 |
+
"source": [
|
| 857 |
+
"# Specify outpit directory and number of checkpoint steps to save\n",
|
| 858 |
+
"run_config = tf.estimator.RunConfig(\n",
|
| 859 |
+
" model_dir=OUTPUT_DIR,\n",
|
| 860 |
+
" save_summary_steps=SAVE_SUMMARY_STEPS,\n",
|
| 861 |
+
" save_checkpoints_steps=SAVE_CHECKPOINTS_STEPS)"
|
| 862 |
+
],
|
| 863 |
+
"execution_count": 0,
|
| 864 |
+
"outputs": []
|
| 865 |
+
},
|
| 866 |
+
{
|
| 867 |
+
"metadata": {
|
| 868 |
+
"id": "q_WebpS1X97v",
|
| 869 |
+
"colab_type": "code",
|
| 870 |
+
"outputId": "1648932a-7391-49d3-8af7-52d514e226e8",
|
| 871 |
+
"colab": {
|
| 872 |
+
"base_uri": "https://localhost:8080/",
|
| 873 |
+
"height": 156
|
| 874 |
+
}
|
| 875 |
+
},
|
| 876 |
+
"cell_type": "code",
|
| 877 |
+
"source": [
|
| 878 |
+
"model_fn = model_fn_builder(\n",
|
| 879 |
+
" num_labels=len(label_list),\n",
|
| 880 |
+
" learning_rate=LEARNING_RATE,\n",
|
| 881 |
+
" num_train_steps=num_train_steps,\n",
|
| 882 |
+
" num_warmup_steps=num_warmup_steps)\n",
|
| 883 |
+
"\n",
|
| 884 |
+
"estimator = tf.estimator.Estimator(\n",
|
| 885 |
+
" model_fn=model_fn,\n",
|
| 886 |
+
" config=run_config,\n",
|
| 887 |
+
" params={\"batch_size\": BATCH_SIZE})\n"
|
| 888 |
+
],
|
| 889 |
+
"execution_count": 55,
|
| 890 |
+
"outputs": [
|
| 891 |
+
{
|
| 892 |
+
"output_type": "stream",
|
| 893 |
+
"text": [
|
| 894 |
+
"INFO:tensorflow:Using config: {'_model_dir': 'gs://bert-tfhub/aclImdb_v1', '_tf_random_seed': None, '_save_summary_steps': 100, '_save_checkpoints_steps': 500, '_save_checkpoints_secs': None, '_session_config': allow_soft_placement: true\n",
|
| 895 |
+
"graph_options {\n",
|
| 896 |
+
" rewrite_options {\n",
|
| 897 |
+
" meta_optimizer_iterations: ONE\n",
|
| 898 |
+
" }\n",
|
| 899 |
+
"}\n",
|
| 900 |
+
", '_keep_checkpoint_max': 5, '_keep_checkpoint_every_n_hours': 10000, '_log_step_count_steps': 100, '_train_distribute': None, '_device_fn': None, '_protocol': None, '_eval_distribute': None, '_experimental_distribute': None, '_service': None, '_cluster_spec': <tensorflow.python.training.server_lib.ClusterSpec object at 0x7fcedb507be0>, '_task_type': 'worker', '_task_id': 0, '_global_id_in_cluster': 0, '_master': '', '_evaluation_master': '', '_is_chief': True, '_num_ps_replicas': 0, '_num_worker_replicas': 1}\n"
|
| 901 |
+
],
|
| 902 |
+
"name": "stdout"
|
| 903 |
+
}
|
| 904 |
+
]
|
| 905 |
+
},
|
| 906 |
+
{
|
| 907 |
+
"metadata": {
|
| 908 |
+
"id": "NOO3RfG1DYLo",
|
| 909 |
+
"colab_type": "text"
|
| 910 |
+
},
|
| 911 |
+
"cell_type": "markdown",
|
| 912 |
+
"source": [
|
| 913 |
+
"Next we create an input builder function that takes our training feature set (`train_features`) and produces a generator. This is a pretty standard design pattern for working with Tensorflow [Estimators](https://www.tensorflow.org/guide/estimators)."
|
| 914 |
+
]
|
| 915 |
+
},
|
| 916 |
+
{
|
| 917 |
+
"metadata": {
|
| 918 |
+
"id": "1Pv2bAlOX_-K",
|
| 919 |
+
"colab_type": "code",
|
| 920 |
+
"colab": {}
|
| 921 |
+
},
|
| 922 |
+
"cell_type": "code",
|
| 923 |
+
"source": [
|
| 924 |
+
"# Create an input function for training. drop_remainder = True for using TPUs.\n",
|
| 925 |
+
"train_input_fn = bert.run_classifier.input_fn_builder(\n",
|
| 926 |
+
" features=train_features,\n",
|
| 927 |
+
" seq_length=MAX_SEQ_LENGTH,\n",
|
| 928 |
+
" is_training=True,\n",
|
| 929 |
+
" drop_remainder=False)"
|
| 930 |
+
],
|
| 931 |
+
"execution_count": 0,
|
| 932 |
+
"outputs": []
|
| 933 |
+
},
|
| 934 |
+
{
|
| 935 |
+
"metadata": {
|
| 936 |
+
"id": "t6Nukby2EB6-",
|
| 937 |
+
"colab_type": "text"
|
| 938 |
+
},
|
| 939 |
+
"cell_type": "markdown",
|
| 940 |
+
"source": [
|
| 941 |
+
"Now we train our model! For me, using a Colab notebook running on Google's GPUs, my training time was about 14 minutes."
|
| 942 |
+
]
|
| 943 |
+
},
|
| 944 |
+
{
|
| 945 |
+
"metadata": {
|
| 946 |
+
"id": "nucD4gluYJmK",
|
| 947 |
+
"colab_type": "code",
|
| 948 |
+
"outputId": "5d728e72-4631-42bf-c48d-3f51d4b968ce",
|
| 949 |
+
"colab": {
|
| 950 |
+
"base_uri": "https://localhost:8080/",
|
| 951 |
+
"height": 68
|
| 952 |
+
}
|
| 953 |
+
},
|
| 954 |
+
"cell_type": "code",
|
| 955 |
+
"source": [
|
| 956 |
+
"print(f'Beginning Training!')\n",
|
| 957 |
+
"current_time = datetime.now()\n",
|
| 958 |
+
"estimator.train(input_fn=train_input_fn, max_steps=num_train_steps)\n",
|
| 959 |
+
"print(\"Training took time \", datetime.now() - current_time)"
|
| 960 |
+
],
|
| 961 |
+
"execution_count": 57,
|
| 962 |
+
"outputs": [
|
| 963 |
+
{
|
| 964 |
+
"output_type": "stream",
|
| 965 |
+
"text": [
|
| 966 |
+
"Beginning Training!\n",
|
| 967 |
+
"INFO:tensorflow:Skipping training since max_steps has already saved.\n",
|
| 968 |
+
"Training took time 0:00:00.759709\n"
|
| 969 |
+
],
|
| 970 |
+
"name": "stdout"
|
| 971 |
+
}
|
| 972 |
+
]
|
| 973 |
+
},
|
| 974 |
+
{
|
| 975 |
+
"metadata": {
|
| 976 |
+
"id": "CmbLTVniARy3",
|
| 977 |
+
"colab_type": "text"
|
| 978 |
+
},
|
| 979 |
+
"cell_type": "markdown",
|
| 980 |
+
"source": [
|
| 981 |
+
"Now let's use our test data to see how well our model did:"
|
| 982 |
+
]
|
| 983 |
+
},
|
| 984 |
+
{
|
| 985 |
+
"metadata": {
|
| 986 |
+
"id": "JIhejfpyJ8Bx",
|
| 987 |
+
"colab_type": "code",
|
| 988 |
+
"colab": {}
|
| 989 |
+
},
|
| 990 |
+
"cell_type": "code",
|
| 991 |
+
"source": [
|
| 992 |
+
"test_input_fn = run_classifier.input_fn_builder(\n",
|
| 993 |
+
" features=test_features,\n",
|
| 994 |
+
" seq_length=MAX_SEQ_LENGTH,\n",
|
| 995 |
+
" is_training=False,\n",
|
| 996 |
+
" drop_remainder=False)"
|
| 997 |
+
],
|
| 998 |
+
"execution_count": 0,
|
| 999 |
+
"outputs": []
|
| 1000 |
+
},
|
| 1001 |
+
{
|
| 1002 |
+
"metadata": {
|
| 1003 |
+
"id": "PPVEXhNjYXC-",
|
| 1004 |
+
"colab_type": "code",
|
| 1005 |
+
"outputId": "dd5482cd-c558-465f-c854-ec11a0175316",
|
| 1006 |
+
"colab": {
|
| 1007 |
+
"base_uri": "https://localhost:8080/",
|
| 1008 |
+
"height": 445
|
| 1009 |
+
}
|
| 1010 |
+
},
|
| 1011 |
+
"cell_type": "code",
|
| 1012 |
+
"source": [
|
| 1013 |
+
"estimator.evaluate(input_fn=test_input_fn, steps=None)"
|
| 1014 |
+
],
|
| 1015 |
+
"execution_count": 59,
|
| 1016 |
+
"outputs": [
|
| 1017 |
+
{
|
| 1018 |
+
"output_type": "stream",
|
| 1019 |
+
"text": [
|
| 1020 |
+
"INFO:tensorflow:Calling model_fn.\n",
|
| 1021 |
+
"INFO:tensorflow:Saver not created because there are no variables in the graph to restore\n"
|
| 1022 |
+
],
|
| 1023 |
+
"name": "stdout"
|
| 1024 |
+
},
|
| 1025 |
+
{
|
| 1026 |
+
"output_type": "stream",
|
| 1027 |
+
"text": [
|
| 1028 |
+
"/usr/local/lib/python3.6/dist-packages/tensorflow/python/ops/gradients_impl.py:110: UserWarning: Converting sparse IndexedSlices to a dense Tensor of unknown shape. This may consume a large amount of memory.\n",
|
| 1029 |
+
" \"Converting sparse IndexedSlices to a dense Tensor of unknown shape. \"\n"
|
| 1030 |
+
],
|
| 1031 |
+
"name": "stderr"
|
| 1032 |
+
},
|
| 1033 |
+
{
|
| 1034 |
+
"output_type": "stream",
|
| 1035 |
+
"text": [
|
| 1036 |
+
"INFO:tensorflow:Done calling model_fn.\n",
|
| 1037 |
+
"INFO:tensorflow:Starting evaluation at 2019-02-12T21:04:20Z\n",
|
| 1038 |
+
"INFO:tensorflow:Graph was finalized.\n",
|
| 1039 |
+
"INFO:tensorflow:Restoring parameters from gs://bert-tfhub/aclImdb_v1/model.ckpt-468\n",
|
| 1040 |
+
"INFO:tensorflow:Running local_init_op.\n",
|
| 1041 |
+
"INFO:tensorflow:Done running local_init_op.\n",
|
| 1042 |
+
"INFO:tensorflow:Finished evaluation at 2019-02-12-21:06:05\n",
|
| 1043 |
+
"INFO:tensorflow:Saving dict for global step 468: auc = 0.86659324, eval_accuracy = 0.8664, f1_score = 0.8659711, false_negatives = 375.0, false_positives = 293.0, global_step = 468, loss = 0.51870537, precision = 0.880457, recall = 0.8519542, true_negatives = 2174.0, true_positives = 2158.0\n",
|
| 1044 |
+
"INFO:tensorflow:Saving 'checkpoint_path' summary for global step 468: gs://bert-tfhub/aclImdb_v1/model.ckpt-468\n"
|
| 1045 |
+
],
|
| 1046 |
+
"name": "stdout"
|
| 1047 |
+
},
|
| 1048 |
+
{
|
| 1049 |
+
"output_type": "execute_result",
|
| 1050 |
+
"data": {
|
| 1051 |
+
"text/plain": [
|
| 1052 |
+
"{'auc': 0.86659324,\n",
|
| 1053 |
+
" 'eval_accuracy': 0.8664,\n",
|
| 1054 |
+
" 'f1_score': 0.8659711,\n",
|
| 1055 |
+
" 'false_negatives': 375.0,\n",
|
| 1056 |
+
" 'false_positives': 293.0,\n",
|
| 1057 |
+
" 'global_step': 468,\n",
|
| 1058 |
+
" 'loss': 0.51870537,\n",
|
| 1059 |
+
" 'precision': 0.880457,\n",
|
| 1060 |
+
" 'recall': 0.8519542,\n",
|
| 1061 |
+
" 'true_negatives': 2174.0,\n",
|
| 1062 |
+
" 'true_positives': 2158.0}"
|
| 1063 |
+
]
|
| 1064 |
+
},
|
| 1065 |
+
"metadata": {
|
| 1066 |
+
"tags": []
|
| 1067 |
+
},
|
| 1068 |
+
"execution_count": 59
|
| 1069 |
+
}
|
| 1070 |
+
]
|
| 1071 |
+
},
|
| 1072 |
+
{
|
| 1073 |
+
"metadata": {
|
| 1074 |
+
"id": "ueKsULteiz1B",
|
| 1075 |
+
"colab_type": "text"
|
| 1076 |
+
},
|
| 1077 |
+
"cell_type": "markdown",
|
| 1078 |
+
"source": [
|
| 1079 |
+
"Now let's write code to make predictions on new sentences:"
|
| 1080 |
+
]
|
| 1081 |
+
},
|
| 1082 |
+
{
|
| 1083 |
+
"metadata": {
|
| 1084 |
+
"id": "OsrbTD2EJTVl",
|
| 1085 |
+
"colab_type": "code",
|
| 1086 |
+
"colab": {}
|
| 1087 |
+
},
|
| 1088 |
+
"cell_type": "code",
|
| 1089 |
+
"source": [
|
| 1090 |
+
"def getPrediction(in_sentences):\n",
|
| 1091 |
+
" labels = [\"Negative\", \"Positive\"]\n",
|
| 1092 |
+
" input_examples = [run_classifier.InputExample(guid=\"\", text_a = x, text_b = None, label = 0) for x in in_sentences] # here, \"\" is just a dummy label\n",
|
| 1093 |
+
" input_features = run_classifier.convert_examples_to_features(input_examples, label_list, MAX_SEQ_LENGTH, tokenizer)\n",
|
| 1094 |
+
" predict_input_fn = run_classifier.input_fn_builder(features=input_features, seq_length=MAX_SEQ_LENGTH, is_training=False, drop_remainder=False)\n",
|
| 1095 |
+
" predictions = estimator.predict(predict_input_fn)\n",
|
| 1096 |
+
" return [(sentence, prediction['probabilities'], labels[prediction['labels']]) for sentence, prediction in zip(in_sentences, predictions)]"
|
| 1097 |
+
],
|
| 1098 |
+
"execution_count": 0,
|
| 1099 |
+
"outputs": []
|
| 1100 |
+
},
|
| 1101 |
+
{
|
| 1102 |
+
"metadata": {
|
| 1103 |
+
"id": "-thbodgih_VJ",
|
| 1104 |
+
"colab_type": "code",
|
| 1105 |
+
"colab": {}
|
| 1106 |
+
},
|
| 1107 |
+
"cell_type": "code",
|
| 1108 |
+
"source": [
|
| 1109 |
+
"pred_sentences = [\n",
|
| 1110 |
+
" \"That movie was absolutely awful\",\n",
|
| 1111 |
+
" \"The acting was a bit lacking\",\n",
|
| 1112 |
+
" \"The film was creative and surprising\",\n",
|
| 1113 |
+
" \"Absolutely fantastic!\"\n",
|
| 1114 |
+
"]"
|
| 1115 |
+
],
|
| 1116 |
+
"execution_count": 0,
|
| 1117 |
+
"outputs": []
|
| 1118 |
+
},
|
| 1119 |
+
{
|
| 1120 |
+
"metadata": {
|
| 1121 |
+
"id": "QrZmvZySKQTm",
|
| 1122 |
+
"colab_type": "code",
|
| 1123 |
+
"colab": {
|
| 1124 |
+
"base_uri": "https://localhost:8080/",
|
| 1125 |
+
"height": 649
|
| 1126 |
+
},
|
| 1127 |
+
"outputId": "3891fafb-a460-4eb8-fa6c-335a5bbc10e5"
|
| 1128 |
+
},
|
| 1129 |
+
"cell_type": "code",
|
| 1130 |
+
"source": [
|
| 1131 |
+
"predictions = getPrediction(pred_sentences)"
|
| 1132 |
+
],
|
| 1133 |
+
"execution_count": 72,
|
| 1134 |
+
"outputs": [
|
| 1135 |
+
{
|
| 1136 |
+
"output_type": "stream",
|
| 1137 |
+
"text": [
|
| 1138 |
+
"INFO:tensorflow:Writing example 0 of 4\n",
|
| 1139 |
+
"INFO:tensorflow:*** Example ***\n",
|
| 1140 |
+
"INFO:tensorflow:guid: \n",
|
| 1141 |
+
"INFO:tensorflow:tokens: [CLS] that movie was absolutely awful [SEP]\n",
|
| 1142 |
+
"INFO:tensorflow:input_ids: 101 2008 3185 2001 7078 9643 102 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n",
|
| 1143 |
+
"INFO:tensorflow:input_mask: 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n",
|
| 1144 |
+
"INFO:tensorflow:segment_ids: 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n",
|
| 1145 |
+
"INFO:tensorflow:label: 0 (id = 0)\n",
|
| 1146 |
+
"INFO:tensorflow:*** Example ***\n",
|
| 1147 |
+
"INFO:tensorflow:guid: \n",
|
| 1148 |
+
"INFO:tensorflow:tokens: [CLS] the acting was a bit lacking [SEP]\n",
|
| 1149 |
+
"INFO:tensorflow:input_ids: 101 1996 3772 2001 1037 2978 11158 102 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n",
|
| 1150 |
+
"INFO:tensorflow:input_mask: 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n",
|
| 1151 |
+
"INFO:tensorflow:segment_ids: 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n",
|
| 1152 |
+
"INFO:tensorflow:label: 0 (id = 0)\n",
|
| 1153 |
+
"INFO:tensorflow:*** Example ***\n",
|
| 1154 |
+
"INFO:tensorflow:guid: \n",
|
| 1155 |
+
"INFO:tensorflow:tokens: [CLS] the film was creative and surprising [SEP]\n",
|
| 1156 |
+
"INFO:tensorflow:input_ids: 101 1996 2143 2001 5541 1998 11341 102 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n",
|
| 1157 |
+
"INFO:tensorflow:input_mask: 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n",
|
| 1158 |
+
"INFO:tensorflow:segment_ids: 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n",
|
| 1159 |
+
"INFO:tensorflow:label: 0 (id = 0)\n",
|
| 1160 |
+
"INFO:tensorflow:*** Example ***\n",
|
| 1161 |
+
"INFO:tensorflow:guid: \n",
|
| 1162 |
+
"INFO:tensorflow:tokens: [CLS] absolutely fantastic ! [SEP]\n",
|
| 1163 |
+
"INFO:tensorflow:input_ids: 101 7078 10392 999 102 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n",
|
| 1164 |
+
"INFO:tensorflow:input_mask: 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n",
|
| 1165 |
+
"INFO:tensorflow:segment_ids: 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0\n",
|
| 1166 |
+
"INFO:tensorflow:label: 0 (id = 0)\n",
|
| 1167 |
+
"INFO:tensorflow:Calling model_fn.\n",
|
| 1168 |
+
"INFO:tensorflow:Saver not created because there are no variables in the graph to restore\n",
|
| 1169 |
+
"INFO:tensorflow:Done calling model_fn.\n",
|
| 1170 |
+
"INFO:tensorflow:Graph was finalized.\n",
|
| 1171 |
+
"INFO:tensorflow:Restoring parameters from gs://bert-tfhub/aclImdb_v1/model.ckpt-468\n",
|
| 1172 |
+
"INFO:tensorflow:Running local_init_op.\n",
|
| 1173 |
+
"INFO:tensorflow:Done running local_init_op.\n"
|
| 1174 |
+
],
|
| 1175 |
+
"name": "stdout"
|
| 1176 |
+
}
|
| 1177 |
+
]
|
| 1178 |
+
},
|
| 1179 |
+
{
|
| 1180 |
+
"metadata": {
|
| 1181 |
+
"id": "MXkRiEBUqN3n",
|
| 1182 |
+
"colab_type": "text"
|
| 1183 |
+
},
|
| 1184 |
+
"cell_type": "markdown",
|
| 1185 |
+
"source": [
|
| 1186 |
+
"Voila! We have a sentiment classifier!"
|
| 1187 |
+
]
|
| 1188 |
+
},
|
| 1189 |
+
{
|
| 1190 |
+
"metadata": {
|
| 1191 |
+
"id": "ERkTE8-7oQLZ",
|
| 1192 |
+
"colab_type": "code",
|
| 1193 |
+
"colab": {
|
| 1194 |
+
"base_uri": "https://localhost:8080/",
|
| 1195 |
+
"height": 221
|
| 1196 |
+
},
|
| 1197 |
+
"outputId": "26c33224-dc2c-4b3d-f7b4-ac3ef0a58b27"
|
| 1198 |
+
},
|
| 1199 |
+
"cell_type": "code",
|
| 1200 |
+
"source": [
|
| 1201 |
+
"predictions"
|
| 1202 |
+
],
|
| 1203 |
+
"execution_count": 73,
|
| 1204 |
+
"outputs": [
|
| 1205 |
+
{
|
| 1206 |
+
"output_type": "execute_result",
|
| 1207 |
+
"data": {
|
| 1208 |
+
"text/plain": [
|
| 1209 |
+
"[('That movie was absolutely awful',\n",
|
| 1210 |
+
" array([-4.9142293e-03, -5.3180690e+00], dtype=float32),\n",
|
| 1211 |
+
" 'Negative'),\n",
|
| 1212 |
+
" ('The acting was a bit lacking',\n",
|
| 1213 |
+
" array([-0.03325794, -3.4200459 ], dtype=float32),\n",
|
| 1214 |
+
" 'Negative'),\n",
|
| 1215 |
+
" ('The film was creative and surprising',\n",
|
| 1216 |
+
" array([-5.3589125e+00, -4.7171740e-03], dtype=float32),\n",
|
| 1217 |
+
" 'Positive'),\n",
|
| 1218 |
+
" ('Absolutely fantastic!',\n",
|
| 1219 |
+
" array([-5.0434084 , -0.00647258], dtype=float32),\n",
|
| 1220 |
+
" 'Positive')]"
|
| 1221 |
+
]
|
| 1222 |
+
},
|
| 1223 |
+
"metadata": {
|
| 1224 |
+
"tags": []
|
| 1225 |
+
},
|
| 1226 |
+
"execution_count": 73
|
| 1227 |
+
}
|
| 1228 |
+
]
|
| 1229 |
+
}
|
| 1230 |
+
]
|
| 1231 |
+
}
|
bert-master/bert-master/requirements.txt
ADDED
|
@@ -0,0 +1,2 @@
|
|
|
|
|
|
|
|
|
|
| 1 |
+
tensorflow >= 1.11.0 # CPU Version of TensorFlow.
|
| 2 |
+
# tensorflow-gpu >= 1.11.0 # GPU version of TensorFlow.
|
bert-master/bert-master/run_classifier.py
ADDED
|
@@ -0,0 +1,981 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# coding=utf-8
|
| 2 |
+
# Copyright 2018 The Google AI Language Team Authors.
|
| 3 |
+
#
|
| 4 |
+
# Licensed under the Apache License, Version 2.0 (the "License");
|
| 5 |
+
# you may not use this file except in compliance with the License.
|
| 6 |
+
# You may obtain a copy of the License at
|
| 7 |
+
#
|
| 8 |
+
# http://www.apache.org/licenses/LICENSE-2.0
|
| 9 |
+
#
|
| 10 |
+
# Unless required by applicable law or agreed to in writing, software
|
| 11 |
+
# distributed under the License is distributed on an "AS IS" BASIS,
|
| 12 |
+
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
| 13 |
+
# See the License for the specific language governing permissions and
|
| 14 |
+
# limitations under the License.
|
| 15 |
+
"""BERT finetuning runner."""
|
| 16 |
+
|
| 17 |
+
from __future__ import absolute_import
|
| 18 |
+
from __future__ import division
|
| 19 |
+
from __future__ import print_function
|
| 20 |
+
|
| 21 |
+
import collections
|
| 22 |
+
import csv
|
| 23 |
+
import os
|
| 24 |
+
import modeling
|
| 25 |
+
import optimization
|
| 26 |
+
import tokenization
|
| 27 |
+
import tensorflow as tf
|
| 28 |
+
|
| 29 |
+
flags = tf.flags
|
| 30 |
+
|
| 31 |
+
FLAGS = flags.FLAGS
|
| 32 |
+
|
| 33 |
+
## Required parameters
|
| 34 |
+
flags.DEFINE_string(
|
| 35 |
+
"data_dir", None,
|
| 36 |
+
"The input data dir. Should contain the .tsv files (or other data files) "
|
| 37 |
+
"for the task.")
|
| 38 |
+
|
| 39 |
+
flags.DEFINE_string(
|
| 40 |
+
"bert_config_file", None,
|
| 41 |
+
"The config json file corresponding to the pre-trained BERT model. "
|
| 42 |
+
"This specifies the model architecture.")
|
| 43 |
+
|
| 44 |
+
flags.DEFINE_string("task_name", None, "The name of the task to train.")
|
| 45 |
+
|
| 46 |
+
flags.DEFINE_string("vocab_file", None,
|
| 47 |
+
"The vocabulary file that the BERT model was trained on.")
|
| 48 |
+
|
| 49 |
+
flags.DEFINE_string(
|
| 50 |
+
"output_dir", None,
|
| 51 |
+
"The output directory where the model checkpoints will be written.")
|
| 52 |
+
|
| 53 |
+
## Other parameters
|
| 54 |
+
|
| 55 |
+
flags.DEFINE_string(
|
| 56 |
+
"init_checkpoint", None,
|
| 57 |
+
"Initial checkpoint (usually from a pre-trained BERT model).")
|
| 58 |
+
|
| 59 |
+
flags.DEFINE_bool(
|
| 60 |
+
"do_lower_case", True,
|
| 61 |
+
"Whether to lower case the input text. Should be True for uncased "
|
| 62 |
+
"models and False for cased models.")
|
| 63 |
+
|
| 64 |
+
flags.DEFINE_integer(
|
| 65 |
+
"max_seq_length", 128,
|
| 66 |
+
"The maximum total input sequence length after WordPiece tokenization. "
|
| 67 |
+
"Sequences longer than this will be truncated, and sequences shorter "
|
| 68 |
+
"than this will be padded.")
|
| 69 |
+
|
| 70 |
+
flags.DEFINE_bool("do_train", False, "Whether to run training.")
|
| 71 |
+
|
| 72 |
+
flags.DEFINE_bool("do_eval", False, "Whether to run eval on the dev set.")
|
| 73 |
+
|
| 74 |
+
flags.DEFINE_bool(
|
| 75 |
+
"do_predict", False,
|
| 76 |
+
"Whether to run the model in inference mode on the test set.")
|
| 77 |
+
|
| 78 |
+
flags.DEFINE_integer("train_batch_size", 32, "Total batch size for training.")
|
| 79 |
+
|
| 80 |
+
flags.DEFINE_integer("eval_batch_size", 8, "Total batch size for eval.")
|
| 81 |
+
|
| 82 |
+
flags.DEFINE_integer("predict_batch_size", 8, "Total batch size for predict.")
|
| 83 |
+
|
| 84 |
+
flags.DEFINE_float("learning_rate", 5e-5, "The initial learning rate for Adam.")
|
| 85 |
+
|
| 86 |
+
flags.DEFINE_float("num_train_epochs", 3.0,
|
| 87 |
+
"Total number of training epochs to perform.")
|
| 88 |
+
|
| 89 |
+
flags.DEFINE_float(
|
| 90 |
+
"warmup_proportion", 0.1,
|
| 91 |
+
"Proportion of training to perform linear learning rate warmup for. "
|
| 92 |
+
"E.g., 0.1 = 10% of training.")
|
| 93 |
+
|
| 94 |
+
flags.DEFINE_integer("save_checkpoints_steps", 1000,
|
| 95 |
+
"How often to save the model checkpoint.")
|
| 96 |
+
|
| 97 |
+
flags.DEFINE_integer("iterations_per_loop", 1000,
|
| 98 |
+
"How many steps to make in each estimator call.")
|
| 99 |
+
|
| 100 |
+
flags.DEFINE_bool("use_tpu", False, "Whether to use TPU or GPU/CPU.")
|
| 101 |
+
|
| 102 |
+
tf.flags.DEFINE_string(
|
| 103 |
+
"tpu_name", None,
|
| 104 |
+
"The Cloud TPU to use for training. This should be either the name "
|
| 105 |
+
"used when creating the Cloud TPU, or a grpc://ip.address.of.tpu:8470 "
|
| 106 |
+
"url.")
|
| 107 |
+
|
| 108 |
+
tf.flags.DEFINE_string(
|
| 109 |
+
"tpu_zone", None,
|
| 110 |
+
"[Optional] GCE zone where the Cloud TPU is located in. If not "
|
| 111 |
+
"specified, we will attempt to automatically detect the GCE project from "
|
| 112 |
+
"metadata.")
|
| 113 |
+
|
| 114 |
+
tf.flags.DEFINE_string(
|
| 115 |
+
"gcp_project", None,
|
| 116 |
+
"[Optional] Project name for the Cloud TPU-enabled project. If not "
|
| 117 |
+
"specified, we will attempt to automatically detect the GCE project from "
|
| 118 |
+
"metadata.")
|
| 119 |
+
|
| 120 |
+
tf.flags.DEFINE_string("master", None, "[Optional] TensorFlow master URL.")
|
| 121 |
+
|
| 122 |
+
flags.DEFINE_integer(
|
| 123 |
+
"num_tpu_cores", 8,
|
| 124 |
+
"Only used if `use_tpu` is True. Total number of TPU cores to use.")
|
| 125 |
+
|
| 126 |
+
|
| 127 |
+
class InputExample(object):
|
| 128 |
+
"""A single training/test example for simple sequence classification."""
|
| 129 |
+
|
| 130 |
+
def __init__(self, guid, text_a, text_b=None, label=None):
|
| 131 |
+
"""Constructs a InputExample.
|
| 132 |
+
|
| 133 |
+
Args:
|
| 134 |
+
guid: Unique id for the example.
|
| 135 |
+
text_a: string. The untokenized text of the first sequence. For single
|
| 136 |
+
sequence tasks, only this sequence must be specified.
|
| 137 |
+
text_b: (Optional) string. The untokenized text of the second sequence.
|
| 138 |
+
Only must be specified for sequence pair tasks.
|
| 139 |
+
label: (Optional) string. The label of the example. This should be
|
| 140 |
+
specified for train and dev examples, but not for test examples.
|
| 141 |
+
"""
|
| 142 |
+
self.guid = guid
|
| 143 |
+
self.text_a = text_a
|
| 144 |
+
self.text_b = text_b
|
| 145 |
+
self.label = label
|
| 146 |
+
|
| 147 |
+
|
| 148 |
+
class PaddingInputExample(object):
|
| 149 |
+
"""Fake example so the num input examples is a multiple of the batch size.
|
| 150 |
+
|
| 151 |
+
When running eval/predict on the TPU, we need to pad the number of examples
|
| 152 |
+
to be a multiple of the batch size, because the TPU requires a fixed batch
|
| 153 |
+
size. The alternative is to drop the last batch, which is bad because it means
|
| 154 |
+
the entire output data won't be generated.
|
| 155 |
+
|
| 156 |
+
We use this class instead of `None` because treating `None` as padding
|
| 157 |
+
battches could cause silent errors.
|
| 158 |
+
"""
|
| 159 |
+
|
| 160 |
+
|
| 161 |
+
class InputFeatures(object):
|
| 162 |
+
"""A single set of features of data."""
|
| 163 |
+
|
| 164 |
+
def __init__(self,
|
| 165 |
+
input_ids,
|
| 166 |
+
input_mask,
|
| 167 |
+
segment_ids,
|
| 168 |
+
label_id,
|
| 169 |
+
is_real_example=True):
|
| 170 |
+
self.input_ids = input_ids
|
| 171 |
+
self.input_mask = input_mask
|
| 172 |
+
self.segment_ids = segment_ids
|
| 173 |
+
self.label_id = label_id
|
| 174 |
+
self.is_real_example = is_real_example
|
| 175 |
+
|
| 176 |
+
|
| 177 |
+
class DataProcessor(object):
|
| 178 |
+
"""Base class for data converters for sequence classification data sets."""
|
| 179 |
+
|
| 180 |
+
def get_train_examples(self, data_dir):
|
| 181 |
+
"""Gets a collection of `InputExample`s for the train set."""
|
| 182 |
+
raise NotImplementedError()
|
| 183 |
+
|
| 184 |
+
def get_dev_examples(self, data_dir):
|
| 185 |
+
"""Gets a collection of `InputExample`s for the dev set."""
|
| 186 |
+
raise NotImplementedError()
|
| 187 |
+
|
| 188 |
+
def get_test_examples(self, data_dir):
|
| 189 |
+
"""Gets a collection of `InputExample`s for prediction."""
|
| 190 |
+
raise NotImplementedError()
|
| 191 |
+
|
| 192 |
+
def get_labels(self):
|
| 193 |
+
"""Gets the list of labels for this data set."""
|
| 194 |
+
raise NotImplementedError()
|
| 195 |
+
|
| 196 |
+
@classmethod
|
| 197 |
+
def _read_tsv(cls, input_file, quotechar=None):
|
| 198 |
+
"""Reads a tab separated value file."""
|
| 199 |
+
with tf.gfile.Open(input_file, "r") as f:
|
| 200 |
+
reader = csv.reader(f, delimiter="\t", quotechar=quotechar)
|
| 201 |
+
lines = []
|
| 202 |
+
for line in reader:
|
| 203 |
+
lines.append(line)
|
| 204 |
+
return lines
|
| 205 |
+
|
| 206 |
+
|
| 207 |
+
class XnliProcessor(DataProcessor):
|
| 208 |
+
"""Processor for the XNLI data set."""
|
| 209 |
+
|
| 210 |
+
def __init__(self):
|
| 211 |
+
self.language = "zh"
|
| 212 |
+
|
| 213 |
+
def get_train_examples(self, data_dir):
|
| 214 |
+
"""See base class."""
|
| 215 |
+
lines = self._read_tsv(
|
| 216 |
+
os.path.join(data_dir, "multinli",
|
| 217 |
+
"multinli.train.%s.tsv" % self.language))
|
| 218 |
+
examples = []
|
| 219 |
+
for (i, line) in enumerate(lines):
|
| 220 |
+
if i == 0:
|
| 221 |
+
continue
|
| 222 |
+
guid = "train-%d" % (i)
|
| 223 |
+
text_a = tokenization.convert_to_unicode(line[0])
|
| 224 |
+
text_b = tokenization.convert_to_unicode(line[1])
|
| 225 |
+
label = tokenization.convert_to_unicode(line[2])
|
| 226 |
+
if label == tokenization.convert_to_unicode("contradictory"):
|
| 227 |
+
label = tokenization.convert_to_unicode("contradiction")
|
| 228 |
+
examples.append(
|
| 229 |
+
InputExample(guid=guid, text_a=text_a, text_b=text_b, label=label))
|
| 230 |
+
return examples
|
| 231 |
+
|
| 232 |
+
def get_dev_examples(self, data_dir):
|
| 233 |
+
"""See base class."""
|
| 234 |
+
lines = self._read_tsv(os.path.join(data_dir, "xnli.dev.tsv"))
|
| 235 |
+
examples = []
|
| 236 |
+
for (i, line) in enumerate(lines):
|
| 237 |
+
if i == 0:
|
| 238 |
+
continue
|
| 239 |
+
guid = "dev-%d" % (i)
|
| 240 |
+
language = tokenization.convert_to_unicode(line[0])
|
| 241 |
+
if language != tokenization.convert_to_unicode(self.language):
|
| 242 |
+
continue
|
| 243 |
+
text_a = tokenization.convert_to_unicode(line[6])
|
| 244 |
+
text_b = tokenization.convert_to_unicode(line[7])
|
| 245 |
+
label = tokenization.convert_to_unicode(line[1])
|
| 246 |
+
examples.append(
|
| 247 |
+
InputExample(guid=guid, text_a=text_a, text_b=text_b, label=label))
|
| 248 |
+
return examples
|
| 249 |
+
|
| 250 |
+
def get_labels(self):
|
| 251 |
+
"""See base class."""
|
| 252 |
+
return ["contradiction", "entailment", "neutral"]
|
| 253 |
+
|
| 254 |
+
|
| 255 |
+
class MnliProcessor(DataProcessor):
|
| 256 |
+
"""Processor for the MultiNLI data set (GLUE version)."""
|
| 257 |
+
|
| 258 |
+
def get_train_examples(self, data_dir):
|
| 259 |
+
"""See base class."""
|
| 260 |
+
return self._create_examples(
|
| 261 |
+
self._read_tsv(os.path.join(data_dir, "train.tsv")), "train")
|
| 262 |
+
|
| 263 |
+
def get_dev_examples(self, data_dir):
|
| 264 |
+
"""See base class."""
|
| 265 |
+
return self._create_examples(
|
| 266 |
+
self._read_tsv(os.path.join(data_dir, "dev_matched.tsv")),
|
| 267 |
+
"dev_matched")
|
| 268 |
+
|
| 269 |
+
def get_test_examples(self, data_dir):
|
| 270 |
+
"""See base class."""
|
| 271 |
+
return self._create_examples(
|
| 272 |
+
self._read_tsv(os.path.join(data_dir, "test_matched.tsv")), "test")
|
| 273 |
+
|
| 274 |
+
def get_labels(self):
|
| 275 |
+
"""See base class."""
|
| 276 |
+
return ["contradiction", "entailment", "neutral"]
|
| 277 |
+
|
| 278 |
+
def _create_examples(self, lines, set_type):
|
| 279 |
+
"""Creates examples for the training and dev sets."""
|
| 280 |
+
examples = []
|
| 281 |
+
for (i, line) in enumerate(lines):
|
| 282 |
+
if i == 0:
|
| 283 |
+
continue
|
| 284 |
+
guid = "%s-%s" % (set_type, tokenization.convert_to_unicode(line[0]))
|
| 285 |
+
text_a = tokenization.convert_to_unicode(line[8])
|
| 286 |
+
text_b = tokenization.convert_to_unicode(line[9])
|
| 287 |
+
if set_type == "test":
|
| 288 |
+
label = "contradiction"
|
| 289 |
+
else:
|
| 290 |
+
label = tokenization.convert_to_unicode(line[-1])
|
| 291 |
+
examples.append(
|
| 292 |
+
InputExample(guid=guid, text_a=text_a, text_b=text_b, label=label))
|
| 293 |
+
return examples
|
| 294 |
+
|
| 295 |
+
|
| 296 |
+
class MrpcProcessor(DataProcessor):
|
| 297 |
+
"""Processor for the MRPC data set (GLUE version)."""
|
| 298 |
+
|
| 299 |
+
def get_train_examples(self, data_dir):
|
| 300 |
+
"""See base class."""
|
| 301 |
+
return self._create_examples(
|
| 302 |
+
self._read_tsv(os.path.join(data_dir, "train.tsv")), "train")
|
| 303 |
+
|
| 304 |
+
def get_dev_examples(self, data_dir):
|
| 305 |
+
"""See base class."""
|
| 306 |
+
return self._create_examples(
|
| 307 |
+
self._read_tsv(os.path.join(data_dir, "dev.tsv")), "dev")
|
| 308 |
+
|
| 309 |
+
def get_test_examples(self, data_dir):
|
| 310 |
+
"""See base class."""
|
| 311 |
+
return self._create_examples(
|
| 312 |
+
self._read_tsv(os.path.join(data_dir, "test.tsv")), "test")
|
| 313 |
+
|
| 314 |
+
def get_labels(self):
|
| 315 |
+
"""See base class."""
|
| 316 |
+
return ["0", "1"]
|
| 317 |
+
|
| 318 |
+
def _create_examples(self, lines, set_type):
|
| 319 |
+
"""Creates examples for the training and dev sets."""
|
| 320 |
+
examples = []
|
| 321 |
+
for (i, line) in enumerate(lines):
|
| 322 |
+
if i == 0:
|
| 323 |
+
continue
|
| 324 |
+
guid = "%s-%s" % (set_type, i)
|
| 325 |
+
text_a = tokenization.convert_to_unicode(line[3])
|
| 326 |
+
text_b = tokenization.convert_to_unicode(line[4])
|
| 327 |
+
if set_type == "test":
|
| 328 |
+
label = "0"
|
| 329 |
+
else:
|
| 330 |
+
label = tokenization.convert_to_unicode(line[0])
|
| 331 |
+
examples.append(
|
| 332 |
+
InputExample(guid=guid, text_a=text_a, text_b=text_b, label=label))
|
| 333 |
+
return examples
|
| 334 |
+
|
| 335 |
+
|
| 336 |
+
class ColaProcessor(DataProcessor):
|
| 337 |
+
"""Processor for the CoLA data set (GLUE version)."""
|
| 338 |
+
|
| 339 |
+
def get_train_examples(self, data_dir):
|
| 340 |
+
"""See base class."""
|
| 341 |
+
return self._create_examples(
|
| 342 |
+
self._read_tsv(os.path.join(data_dir, "train.tsv")), "train")
|
| 343 |
+
|
| 344 |
+
def get_dev_examples(self, data_dir):
|
| 345 |
+
"""See base class."""
|
| 346 |
+
return self._create_examples(
|
| 347 |
+
self._read_tsv(os.path.join(data_dir, "dev.tsv")), "dev")
|
| 348 |
+
|
| 349 |
+
def get_test_examples(self, data_dir):
|
| 350 |
+
"""See base class."""
|
| 351 |
+
return self._create_examples(
|
| 352 |
+
self._read_tsv(os.path.join(data_dir, "test.tsv")), "test")
|
| 353 |
+
|
| 354 |
+
def get_labels(self):
|
| 355 |
+
"""See base class."""
|
| 356 |
+
return ["0", "1"]
|
| 357 |
+
|
| 358 |
+
def _create_examples(self, lines, set_type):
|
| 359 |
+
"""Creates examples for the training and dev sets."""
|
| 360 |
+
examples = []
|
| 361 |
+
for (i, line) in enumerate(lines):
|
| 362 |
+
# Only the test set has a header
|
| 363 |
+
if set_type == "test" and i == 0:
|
| 364 |
+
continue
|
| 365 |
+
guid = "%s-%s" % (set_type, i)
|
| 366 |
+
if set_type == "test":
|
| 367 |
+
text_a = tokenization.convert_to_unicode(line[1])
|
| 368 |
+
label = "0"
|
| 369 |
+
else:
|
| 370 |
+
text_a = tokenization.convert_to_unicode(line[3])
|
| 371 |
+
label = tokenization.convert_to_unicode(line[1])
|
| 372 |
+
examples.append(
|
| 373 |
+
InputExample(guid=guid, text_a=text_a, text_b=None, label=label))
|
| 374 |
+
return examples
|
| 375 |
+
|
| 376 |
+
|
| 377 |
+
def convert_single_example(ex_index, example, label_list, max_seq_length,
|
| 378 |
+
tokenizer):
|
| 379 |
+
"""Converts a single `InputExample` into a single `InputFeatures`."""
|
| 380 |
+
|
| 381 |
+
if isinstance(example, PaddingInputExample):
|
| 382 |
+
return InputFeatures(
|
| 383 |
+
input_ids=[0] * max_seq_length,
|
| 384 |
+
input_mask=[0] * max_seq_length,
|
| 385 |
+
segment_ids=[0] * max_seq_length,
|
| 386 |
+
label_id=0,
|
| 387 |
+
is_real_example=False)
|
| 388 |
+
|
| 389 |
+
label_map = {}
|
| 390 |
+
for (i, label) in enumerate(label_list):
|
| 391 |
+
label_map[label] = i
|
| 392 |
+
|
| 393 |
+
tokens_a = tokenizer.tokenize(example.text_a)
|
| 394 |
+
tokens_b = None
|
| 395 |
+
if example.text_b:
|
| 396 |
+
tokens_b = tokenizer.tokenize(example.text_b)
|
| 397 |
+
|
| 398 |
+
if tokens_b:
|
| 399 |
+
# Modifies `tokens_a` and `tokens_b` in place so that the total
|
| 400 |
+
# length is less than the specified length.
|
| 401 |
+
# Account for [CLS], [SEP], [SEP] with "- 3"
|
| 402 |
+
_truncate_seq_pair(tokens_a, tokens_b, max_seq_length - 3)
|
| 403 |
+
else:
|
| 404 |
+
# Account for [CLS] and [SEP] with "- 2"
|
| 405 |
+
if len(tokens_a) > max_seq_length - 2:
|
| 406 |
+
tokens_a = tokens_a[0:(max_seq_length - 2)]
|
| 407 |
+
|
| 408 |
+
# The convention in BERT is:
|
| 409 |
+
# (a) For sequence pairs:
|
| 410 |
+
# tokens: [CLS] is this jack ##son ##ville ? [SEP] no it is not . [SEP]
|
| 411 |
+
# type_ids: 0 0 0 0 0 0 0 0 1 1 1 1 1 1
|
| 412 |
+
# (b) For single sequences:
|
| 413 |
+
# tokens: [CLS] the dog is hairy . [SEP]
|
| 414 |
+
# type_ids: 0 0 0 0 0 0 0
|
| 415 |
+
#
|
| 416 |
+
# Where "type_ids" are used to indicate whether this is the first
|
| 417 |
+
# sequence or the second sequence. The embedding vectors for `type=0` and
|
| 418 |
+
# `type=1` were learned during pre-training and are added to the wordpiece
|
| 419 |
+
# embedding vector (and position vector). This is not *strictly* necessary
|
| 420 |
+
# since the [SEP] token unambiguously separates the sequences, but it makes
|
| 421 |
+
# it easier for the model to learn the concept of sequences.
|
| 422 |
+
#
|
| 423 |
+
# For classification tasks, the first vector (corresponding to [CLS]) is
|
| 424 |
+
# used as the "sentence vector". Note that this only makes sense because
|
| 425 |
+
# the entire model is fine-tuned.
|
| 426 |
+
tokens = []
|
| 427 |
+
segment_ids = []
|
| 428 |
+
tokens.append("[CLS]")
|
| 429 |
+
segment_ids.append(0)
|
| 430 |
+
for token in tokens_a:
|
| 431 |
+
tokens.append(token)
|
| 432 |
+
segment_ids.append(0)
|
| 433 |
+
tokens.append("[SEP]")
|
| 434 |
+
segment_ids.append(0)
|
| 435 |
+
|
| 436 |
+
if tokens_b:
|
| 437 |
+
for token in tokens_b:
|
| 438 |
+
tokens.append(token)
|
| 439 |
+
segment_ids.append(1)
|
| 440 |
+
tokens.append("[SEP]")
|
| 441 |
+
segment_ids.append(1)
|
| 442 |
+
|
| 443 |
+
input_ids = tokenizer.convert_tokens_to_ids(tokens)
|
| 444 |
+
|
| 445 |
+
# The mask has 1 for real tokens and 0 for padding tokens. Only real
|
| 446 |
+
# tokens are attended to.
|
| 447 |
+
input_mask = [1] * len(input_ids)
|
| 448 |
+
|
| 449 |
+
# Zero-pad up to the sequence length.
|
| 450 |
+
while len(input_ids) < max_seq_length:
|
| 451 |
+
input_ids.append(0)
|
| 452 |
+
input_mask.append(0)
|
| 453 |
+
segment_ids.append(0)
|
| 454 |
+
|
| 455 |
+
assert len(input_ids) == max_seq_length
|
| 456 |
+
assert len(input_mask) == max_seq_length
|
| 457 |
+
assert len(segment_ids) == max_seq_length
|
| 458 |
+
|
| 459 |
+
label_id = label_map[example.label]
|
| 460 |
+
if ex_index < 5:
|
| 461 |
+
tf.logging.info("*** Example ***")
|
| 462 |
+
tf.logging.info("guid: %s" % (example.guid))
|
| 463 |
+
tf.logging.info("tokens: %s" % " ".join(
|
| 464 |
+
[tokenization.printable_text(x) for x in tokens]))
|
| 465 |
+
tf.logging.info("input_ids: %s" % " ".join([str(x) for x in input_ids]))
|
| 466 |
+
tf.logging.info("input_mask: %s" % " ".join([str(x) for x in input_mask]))
|
| 467 |
+
tf.logging.info("segment_ids: %s" % " ".join([str(x) for x in segment_ids]))
|
| 468 |
+
tf.logging.info("label: %s (id = %d)" % (example.label, label_id))
|
| 469 |
+
|
| 470 |
+
feature = InputFeatures(
|
| 471 |
+
input_ids=input_ids,
|
| 472 |
+
input_mask=input_mask,
|
| 473 |
+
segment_ids=segment_ids,
|
| 474 |
+
label_id=label_id,
|
| 475 |
+
is_real_example=True)
|
| 476 |
+
return feature
|
| 477 |
+
|
| 478 |
+
|
| 479 |
+
def file_based_convert_examples_to_features(
|
| 480 |
+
examples, label_list, max_seq_length, tokenizer, output_file):
|
| 481 |
+
"""Convert a set of `InputExample`s to a TFRecord file."""
|
| 482 |
+
|
| 483 |
+
writer = tf.python_io.TFRecordWriter(output_file)
|
| 484 |
+
|
| 485 |
+
for (ex_index, example) in enumerate(examples):
|
| 486 |
+
if ex_index % 10000 == 0:
|
| 487 |
+
tf.logging.info("Writing example %d of %d" % (ex_index, len(examples)))
|
| 488 |
+
|
| 489 |
+
feature = convert_single_example(ex_index, example, label_list,
|
| 490 |
+
max_seq_length, tokenizer)
|
| 491 |
+
|
| 492 |
+
def create_int_feature(values):
|
| 493 |
+
f = tf.train.Feature(int64_list=tf.train.Int64List(value=list(values)))
|
| 494 |
+
return f
|
| 495 |
+
|
| 496 |
+
features = collections.OrderedDict()
|
| 497 |
+
features["input_ids"] = create_int_feature(feature.input_ids)
|
| 498 |
+
features["input_mask"] = create_int_feature(feature.input_mask)
|
| 499 |
+
features["segment_ids"] = create_int_feature(feature.segment_ids)
|
| 500 |
+
features["label_ids"] = create_int_feature([feature.label_id])
|
| 501 |
+
features["is_real_example"] = create_int_feature(
|
| 502 |
+
[int(feature.is_real_example)])
|
| 503 |
+
|
| 504 |
+
tf_example = tf.train.Example(features=tf.train.Features(feature=features))
|
| 505 |
+
writer.write(tf_example.SerializeToString())
|
| 506 |
+
writer.close()
|
| 507 |
+
|
| 508 |
+
|
| 509 |
+
def file_based_input_fn_builder(input_file, seq_length, is_training,
|
| 510 |
+
drop_remainder):
|
| 511 |
+
"""Creates an `input_fn` closure to be passed to TPUEstimator."""
|
| 512 |
+
|
| 513 |
+
name_to_features = {
|
| 514 |
+
"input_ids": tf.FixedLenFeature([seq_length], tf.int64),
|
| 515 |
+
"input_mask": tf.FixedLenFeature([seq_length], tf.int64),
|
| 516 |
+
"segment_ids": tf.FixedLenFeature([seq_length], tf.int64),
|
| 517 |
+
"label_ids": tf.FixedLenFeature([], tf.int64),
|
| 518 |
+
"is_real_example": tf.FixedLenFeature([], tf.int64),
|
| 519 |
+
}
|
| 520 |
+
|
| 521 |
+
def _decode_record(record, name_to_features):
|
| 522 |
+
"""Decodes a record to a TensorFlow example."""
|
| 523 |
+
example = tf.parse_single_example(record, name_to_features)
|
| 524 |
+
|
| 525 |
+
# tf.Example only supports tf.int64, but the TPU only supports tf.int32.
|
| 526 |
+
# So cast all int64 to int32.
|
| 527 |
+
for name in list(example.keys()):
|
| 528 |
+
t = example[name]
|
| 529 |
+
if t.dtype == tf.int64:
|
| 530 |
+
t = tf.to_int32(t)
|
| 531 |
+
example[name] = t
|
| 532 |
+
|
| 533 |
+
return example
|
| 534 |
+
|
| 535 |
+
def input_fn(params):
|
| 536 |
+
"""The actual input function."""
|
| 537 |
+
batch_size = params["batch_size"]
|
| 538 |
+
|
| 539 |
+
# For training, we want a lot of parallel reading and shuffling.
|
| 540 |
+
# For eval, we want no shuffling and parallel reading doesn't matter.
|
| 541 |
+
d = tf.data.TFRecordDataset(input_file)
|
| 542 |
+
if is_training:
|
| 543 |
+
d = d.repeat()
|
| 544 |
+
d = d.shuffle(buffer_size=100)
|
| 545 |
+
|
| 546 |
+
d = d.apply(
|
| 547 |
+
tf.contrib.data.map_and_batch(
|
| 548 |
+
lambda record: _decode_record(record, name_to_features),
|
| 549 |
+
batch_size=batch_size,
|
| 550 |
+
drop_remainder=drop_remainder))
|
| 551 |
+
|
| 552 |
+
return d
|
| 553 |
+
|
| 554 |
+
return input_fn
|
| 555 |
+
|
| 556 |
+
|
| 557 |
+
def _truncate_seq_pair(tokens_a, tokens_b, max_length):
|
| 558 |
+
"""Truncates a sequence pair in place to the maximum length."""
|
| 559 |
+
|
| 560 |
+
# This is a simple heuristic which will always truncate the longer sequence
|
| 561 |
+
# one token at a time. This makes more sense than truncating an equal percent
|
| 562 |
+
# of tokens from each, since if one sequence is very short then each token
|
| 563 |
+
# that's truncated likely contains more information than a longer sequence.
|
| 564 |
+
while True:
|
| 565 |
+
total_length = len(tokens_a) + len(tokens_b)
|
| 566 |
+
if total_length <= max_length:
|
| 567 |
+
break
|
| 568 |
+
if len(tokens_a) > len(tokens_b):
|
| 569 |
+
tokens_a.pop()
|
| 570 |
+
else:
|
| 571 |
+
tokens_b.pop()
|
| 572 |
+
|
| 573 |
+
|
| 574 |
+
def create_model(bert_config, is_training, input_ids, input_mask, segment_ids,
|
| 575 |
+
labels, num_labels, use_one_hot_embeddings):
|
| 576 |
+
"""Creates a classification model."""
|
| 577 |
+
model = modeling.BertModel(
|
| 578 |
+
config=bert_config,
|
| 579 |
+
is_training=is_training,
|
| 580 |
+
input_ids=input_ids,
|
| 581 |
+
input_mask=input_mask,
|
| 582 |
+
token_type_ids=segment_ids,
|
| 583 |
+
use_one_hot_embeddings=use_one_hot_embeddings)
|
| 584 |
+
|
| 585 |
+
# In the demo, we are doing a simple classification task on the entire
|
| 586 |
+
# segment.
|
| 587 |
+
#
|
| 588 |
+
# If you want to use the token-level output, use model.get_sequence_output()
|
| 589 |
+
# instead.
|
| 590 |
+
output_layer = model.get_pooled_output()
|
| 591 |
+
|
| 592 |
+
hidden_size = output_layer.shape[-1].value
|
| 593 |
+
|
| 594 |
+
output_weights = tf.get_variable(
|
| 595 |
+
"output_weights", [num_labels, hidden_size],
|
| 596 |
+
initializer=tf.truncated_normal_initializer(stddev=0.02))
|
| 597 |
+
|
| 598 |
+
output_bias = tf.get_variable(
|
| 599 |
+
"output_bias", [num_labels], initializer=tf.zeros_initializer())
|
| 600 |
+
|
| 601 |
+
with tf.variable_scope("loss"):
|
| 602 |
+
if is_training:
|
| 603 |
+
# I.e., 0.1 dropout
|
| 604 |
+
output_layer = tf.nn.dropout(output_layer, keep_prob=0.9)
|
| 605 |
+
|
| 606 |
+
logits = tf.matmul(output_layer, output_weights, transpose_b=True)
|
| 607 |
+
logits = tf.nn.bias_add(logits, output_bias)
|
| 608 |
+
probabilities = tf.nn.softmax(logits, axis=-1)
|
| 609 |
+
log_probs = tf.nn.log_softmax(logits, axis=-1)
|
| 610 |
+
|
| 611 |
+
one_hot_labels = tf.one_hot(labels, depth=num_labels, dtype=tf.float32)
|
| 612 |
+
|
| 613 |
+
per_example_loss = -tf.reduce_sum(one_hot_labels * log_probs, axis=-1)
|
| 614 |
+
loss = tf.reduce_mean(per_example_loss)
|
| 615 |
+
|
| 616 |
+
return (loss, per_example_loss, logits, probabilities)
|
| 617 |
+
|
| 618 |
+
|
| 619 |
+
def model_fn_builder(bert_config, num_labels, init_checkpoint, learning_rate,
|
| 620 |
+
num_train_steps, num_warmup_steps, use_tpu,
|
| 621 |
+
use_one_hot_embeddings):
|
| 622 |
+
"""Returns `model_fn` closure for TPUEstimator."""
|
| 623 |
+
|
| 624 |
+
def model_fn(features, labels, mode, params): # pylint: disable=unused-argument
|
| 625 |
+
"""The `model_fn` for TPUEstimator."""
|
| 626 |
+
|
| 627 |
+
tf.logging.info("*** Features ***")
|
| 628 |
+
for name in sorted(features.keys()):
|
| 629 |
+
tf.logging.info(" name = %s, shape = %s" % (name, features[name].shape))
|
| 630 |
+
|
| 631 |
+
input_ids = features["input_ids"]
|
| 632 |
+
input_mask = features["input_mask"]
|
| 633 |
+
segment_ids = features["segment_ids"]
|
| 634 |
+
label_ids = features["label_ids"]
|
| 635 |
+
is_real_example = None
|
| 636 |
+
if "is_real_example" in features:
|
| 637 |
+
is_real_example = tf.cast(features["is_real_example"], dtype=tf.float32)
|
| 638 |
+
else:
|
| 639 |
+
is_real_example = tf.ones(tf.shape(label_ids), dtype=tf.float32)
|
| 640 |
+
|
| 641 |
+
is_training = (mode == tf.estimator.ModeKeys.TRAIN)
|
| 642 |
+
|
| 643 |
+
(total_loss, per_example_loss, logits, probabilities) = create_model(
|
| 644 |
+
bert_config, is_training, input_ids, input_mask, segment_ids, label_ids,
|
| 645 |
+
num_labels, use_one_hot_embeddings)
|
| 646 |
+
|
| 647 |
+
tvars = tf.trainable_variables()
|
| 648 |
+
initialized_variable_names = {}
|
| 649 |
+
scaffold_fn = None
|
| 650 |
+
if init_checkpoint:
|
| 651 |
+
(assignment_map, initialized_variable_names
|
| 652 |
+
) = modeling.get_assignment_map_from_checkpoint(tvars, init_checkpoint)
|
| 653 |
+
if use_tpu:
|
| 654 |
+
|
| 655 |
+
def tpu_scaffold():
|
| 656 |
+
tf.train.init_from_checkpoint(init_checkpoint, assignment_map)
|
| 657 |
+
return tf.train.Scaffold()
|
| 658 |
+
|
| 659 |
+
scaffold_fn = tpu_scaffold
|
| 660 |
+
else:
|
| 661 |
+
tf.train.init_from_checkpoint(init_checkpoint, assignment_map)
|
| 662 |
+
|
| 663 |
+
tf.logging.info("**** Trainable Variables ****")
|
| 664 |
+
for var in tvars:
|
| 665 |
+
init_string = ""
|
| 666 |
+
if var.name in initialized_variable_names:
|
| 667 |
+
init_string = ", *INIT_FROM_CKPT*"
|
| 668 |
+
tf.logging.info(" name = %s, shape = %s%s", var.name, var.shape,
|
| 669 |
+
init_string)
|
| 670 |
+
|
| 671 |
+
output_spec = None
|
| 672 |
+
if mode == tf.estimator.ModeKeys.TRAIN:
|
| 673 |
+
|
| 674 |
+
train_op = optimization.create_optimizer(
|
| 675 |
+
total_loss, learning_rate, num_train_steps, num_warmup_steps, use_tpu)
|
| 676 |
+
|
| 677 |
+
output_spec = tf.contrib.tpu.TPUEstimatorSpec(
|
| 678 |
+
mode=mode,
|
| 679 |
+
loss=total_loss,
|
| 680 |
+
train_op=train_op,
|
| 681 |
+
scaffold_fn=scaffold_fn)
|
| 682 |
+
elif mode == tf.estimator.ModeKeys.EVAL:
|
| 683 |
+
|
| 684 |
+
def metric_fn(per_example_loss, label_ids, logits, is_real_example):
|
| 685 |
+
predictions = tf.argmax(logits, axis=-1, output_type=tf.int32)
|
| 686 |
+
accuracy = tf.metrics.accuracy(
|
| 687 |
+
labels=label_ids, predictions=predictions, weights=is_real_example)
|
| 688 |
+
loss = tf.metrics.mean(values=per_example_loss, weights=is_real_example)
|
| 689 |
+
return {
|
| 690 |
+
"eval_accuracy": accuracy,
|
| 691 |
+
"eval_loss": loss,
|
| 692 |
+
}
|
| 693 |
+
|
| 694 |
+
eval_metrics = (metric_fn,
|
| 695 |
+
[per_example_loss, label_ids, logits, is_real_example])
|
| 696 |
+
output_spec = tf.contrib.tpu.TPUEstimatorSpec(
|
| 697 |
+
mode=mode,
|
| 698 |
+
loss=total_loss,
|
| 699 |
+
eval_metrics=eval_metrics,
|
| 700 |
+
scaffold_fn=scaffold_fn)
|
| 701 |
+
else:
|
| 702 |
+
output_spec = tf.contrib.tpu.TPUEstimatorSpec(
|
| 703 |
+
mode=mode,
|
| 704 |
+
predictions={"probabilities": probabilities},
|
| 705 |
+
scaffold_fn=scaffold_fn)
|
| 706 |
+
return output_spec
|
| 707 |
+
|
| 708 |
+
return model_fn
|
| 709 |
+
|
| 710 |
+
|
| 711 |
+
# This function is not used by this file but is still used by the Colab and
|
| 712 |
+
# people who depend on it.
|
| 713 |
+
def input_fn_builder(features, seq_length, is_training, drop_remainder):
|
| 714 |
+
"""Creates an `input_fn` closure to be passed to TPUEstimator."""
|
| 715 |
+
|
| 716 |
+
all_input_ids = []
|
| 717 |
+
all_input_mask = []
|
| 718 |
+
all_segment_ids = []
|
| 719 |
+
all_label_ids = []
|
| 720 |
+
|
| 721 |
+
for feature in features:
|
| 722 |
+
all_input_ids.append(feature.input_ids)
|
| 723 |
+
all_input_mask.append(feature.input_mask)
|
| 724 |
+
all_segment_ids.append(feature.segment_ids)
|
| 725 |
+
all_label_ids.append(feature.label_id)
|
| 726 |
+
|
| 727 |
+
def input_fn(params):
|
| 728 |
+
"""The actual input function."""
|
| 729 |
+
batch_size = params["batch_size"]
|
| 730 |
+
|
| 731 |
+
num_examples = len(features)
|
| 732 |
+
|
| 733 |
+
# This is for demo purposes and does NOT scale to large data sets. We do
|
| 734 |
+
# not use Dataset.from_generator() because that uses tf.py_func which is
|
| 735 |
+
# not TPU compatible. The right way to load data is with TFRecordReader.
|
| 736 |
+
d = tf.data.Dataset.from_tensor_slices({
|
| 737 |
+
"input_ids":
|
| 738 |
+
tf.constant(
|
| 739 |
+
all_input_ids, shape=[num_examples, seq_length],
|
| 740 |
+
dtype=tf.int32),
|
| 741 |
+
"input_mask":
|
| 742 |
+
tf.constant(
|
| 743 |
+
all_input_mask,
|
| 744 |
+
shape=[num_examples, seq_length],
|
| 745 |
+
dtype=tf.int32),
|
| 746 |
+
"segment_ids":
|
| 747 |
+
tf.constant(
|
| 748 |
+
all_segment_ids,
|
| 749 |
+
shape=[num_examples, seq_length],
|
| 750 |
+
dtype=tf.int32),
|
| 751 |
+
"label_ids":
|
| 752 |
+
tf.constant(all_label_ids, shape=[num_examples], dtype=tf.int32),
|
| 753 |
+
})
|
| 754 |
+
|
| 755 |
+
if is_training:
|
| 756 |
+
d = d.repeat()
|
| 757 |
+
d = d.shuffle(buffer_size=100)
|
| 758 |
+
|
| 759 |
+
d = d.batch(batch_size=batch_size, drop_remainder=drop_remainder)
|
| 760 |
+
return d
|
| 761 |
+
|
| 762 |
+
return input_fn
|
| 763 |
+
|
| 764 |
+
|
| 765 |
+
# This function is not used by this file but is still used by the Colab and
|
| 766 |
+
# people who depend on it.
|
| 767 |
+
def convert_examples_to_features(examples, label_list, max_seq_length,
|
| 768 |
+
tokenizer):
|
| 769 |
+
"""Convert a set of `InputExample`s to a list of `InputFeatures`."""
|
| 770 |
+
|
| 771 |
+
features = []
|
| 772 |
+
for (ex_index, example) in enumerate(examples):
|
| 773 |
+
if ex_index % 10000 == 0:
|
| 774 |
+
tf.logging.info("Writing example %d of %d" % (ex_index, len(examples)))
|
| 775 |
+
|
| 776 |
+
feature = convert_single_example(ex_index, example, label_list,
|
| 777 |
+
max_seq_length, tokenizer)
|
| 778 |
+
|
| 779 |
+
features.append(feature)
|
| 780 |
+
return features
|
| 781 |
+
|
| 782 |
+
|
| 783 |
+
def main(_):
|
| 784 |
+
tf.logging.set_verbosity(tf.logging.INFO)
|
| 785 |
+
|
| 786 |
+
processors = {
|
| 787 |
+
"cola": ColaProcessor,
|
| 788 |
+
"mnli": MnliProcessor,
|
| 789 |
+
"mrpc": MrpcProcessor,
|
| 790 |
+
"xnli": XnliProcessor,
|
| 791 |
+
}
|
| 792 |
+
|
| 793 |
+
tokenization.validate_case_matches_checkpoint(FLAGS.do_lower_case,
|
| 794 |
+
FLAGS.init_checkpoint)
|
| 795 |
+
|
| 796 |
+
if not FLAGS.do_train and not FLAGS.do_eval and not FLAGS.do_predict:
|
| 797 |
+
raise ValueError(
|
| 798 |
+
"At least one of `do_train`, `do_eval` or `do_predict' must be True.")
|
| 799 |
+
|
| 800 |
+
bert_config = modeling.BertConfig.from_json_file(FLAGS.bert_config_file)
|
| 801 |
+
|
| 802 |
+
if FLAGS.max_seq_length > bert_config.max_position_embeddings:
|
| 803 |
+
raise ValueError(
|
| 804 |
+
"Cannot use sequence length %d because the BERT model "
|
| 805 |
+
"was only trained up to sequence length %d" %
|
| 806 |
+
(FLAGS.max_seq_length, bert_config.max_position_embeddings))
|
| 807 |
+
|
| 808 |
+
tf.gfile.MakeDirs(FLAGS.output_dir)
|
| 809 |
+
|
| 810 |
+
task_name = FLAGS.task_name.lower()
|
| 811 |
+
|
| 812 |
+
if task_name not in processors:
|
| 813 |
+
raise ValueError("Task not found: %s" % (task_name))
|
| 814 |
+
|
| 815 |
+
processor = processors[task_name]()
|
| 816 |
+
|
| 817 |
+
label_list = processor.get_labels()
|
| 818 |
+
|
| 819 |
+
tokenizer = tokenization.FullTokenizer(
|
| 820 |
+
vocab_file=FLAGS.vocab_file, do_lower_case=FLAGS.do_lower_case)
|
| 821 |
+
|
| 822 |
+
tpu_cluster_resolver = None
|
| 823 |
+
if FLAGS.use_tpu and FLAGS.tpu_name:
|
| 824 |
+
tpu_cluster_resolver = tf.contrib.cluster_resolver.TPUClusterResolver(
|
| 825 |
+
FLAGS.tpu_name, zone=FLAGS.tpu_zone, project=FLAGS.gcp_project)
|
| 826 |
+
|
| 827 |
+
is_per_host = tf.contrib.tpu.InputPipelineConfig.PER_HOST_V2
|
| 828 |
+
run_config = tf.contrib.tpu.RunConfig(
|
| 829 |
+
cluster=tpu_cluster_resolver,
|
| 830 |
+
master=FLAGS.master,
|
| 831 |
+
model_dir=FLAGS.output_dir,
|
| 832 |
+
save_checkpoints_steps=FLAGS.save_checkpoints_steps,
|
| 833 |
+
tpu_config=tf.contrib.tpu.TPUConfig(
|
| 834 |
+
iterations_per_loop=FLAGS.iterations_per_loop,
|
| 835 |
+
num_shards=FLAGS.num_tpu_cores,
|
| 836 |
+
per_host_input_for_training=is_per_host))
|
| 837 |
+
|
| 838 |
+
train_examples = None
|
| 839 |
+
num_train_steps = None
|
| 840 |
+
num_warmup_steps = None
|
| 841 |
+
if FLAGS.do_train:
|
| 842 |
+
train_examples = processor.get_train_examples(FLAGS.data_dir)
|
| 843 |
+
num_train_steps = int(
|
| 844 |
+
len(train_examples) / FLAGS.train_batch_size * FLAGS.num_train_epochs)
|
| 845 |
+
num_warmup_steps = int(num_train_steps * FLAGS.warmup_proportion)
|
| 846 |
+
|
| 847 |
+
model_fn = model_fn_builder(
|
| 848 |
+
bert_config=bert_config,
|
| 849 |
+
num_labels=len(label_list),
|
| 850 |
+
init_checkpoint=FLAGS.init_checkpoint,
|
| 851 |
+
learning_rate=FLAGS.learning_rate,
|
| 852 |
+
num_train_steps=num_train_steps,
|
| 853 |
+
num_warmup_steps=num_warmup_steps,
|
| 854 |
+
use_tpu=FLAGS.use_tpu,
|
| 855 |
+
use_one_hot_embeddings=FLAGS.use_tpu)
|
| 856 |
+
|
| 857 |
+
# If TPU is not available, this will fall back to normal Estimator on CPU
|
| 858 |
+
# or GPU.
|
| 859 |
+
estimator = tf.contrib.tpu.TPUEstimator(
|
| 860 |
+
use_tpu=FLAGS.use_tpu,
|
| 861 |
+
model_fn=model_fn,
|
| 862 |
+
config=run_config,
|
| 863 |
+
train_batch_size=FLAGS.train_batch_size,
|
| 864 |
+
eval_batch_size=FLAGS.eval_batch_size,
|
| 865 |
+
predict_batch_size=FLAGS.predict_batch_size)
|
| 866 |
+
|
| 867 |
+
if FLAGS.do_train:
|
| 868 |
+
train_file = os.path.join(FLAGS.output_dir, "train.tf_record")
|
| 869 |
+
file_based_convert_examples_to_features(
|
| 870 |
+
train_examples, label_list, FLAGS.max_seq_length, tokenizer, train_file)
|
| 871 |
+
tf.logging.info("***** Running training *****")
|
| 872 |
+
tf.logging.info(" Num examples = %d", len(train_examples))
|
| 873 |
+
tf.logging.info(" Batch size = %d", FLAGS.train_batch_size)
|
| 874 |
+
tf.logging.info(" Num steps = %d", num_train_steps)
|
| 875 |
+
train_input_fn = file_based_input_fn_builder(
|
| 876 |
+
input_file=train_file,
|
| 877 |
+
seq_length=FLAGS.max_seq_length,
|
| 878 |
+
is_training=True,
|
| 879 |
+
drop_remainder=True)
|
| 880 |
+
estimator.train(input_fn=train_input_fn, max_steps=num_train_steps)
|
| 881 |
+
|
| 882 |
+
if FLAGS.do_eval:
|
| 883 |
+
eval_examples = processor.get_dev_examples(FLAGS.data_dir)
|
| 884 |
+
num_actual_eval_examples = len(eval_examples)
|
| 885 |
+
if FLAGS.use_tpu:
|
| 886 |
+
# TPU requires a fixed batch size for all batches, therefore the number
|
| 887 |
+
# of examples must be a multiple of the batch size, or else examples
|
| 888 |
+
# will get dropped. So we pad with fake examples which are ignored
|
| 889 |
+
# later on. These do NOT count towards the metric (all tf.metrics
|
| 890 |
+
# support a per-instance weight, and these get a weight of 0.0).
|
| 891 |
+
while len(eval_examples) % FLAGS.eval_batch_size != 0:
|
| 892 |
+
eval_examples.append(PaddingInputExample())
|
| 893 |
+
|
| 894 |
+
eval_file = os.path.join(FLAGS.output_dir, "eval.tf_record")
|
| 895 |
+
file_based_convert_examples_to_features(
|
| 896 |
+
eval_examples, label_list, FLAGS.max_seq_length, tokenizer, eval_file)
|
| 897 |
+
|
| 898 |
+
tf.logging.info("***** Running evaluation *****")
|
| 899 |
+
tf.logging.info(" Num examples = %d (%d actual, %d padding)",
|
| 900 |
+
len(eval_examples), num_actual_eval_examples,
|
| 901 |
+
len(eval_examples) - num_actual_eval_examples)
|
| 902 |
+
tf.logging.info(" Batch size = %d", FLAGS.eval_batch_size)
|
| 903 |
+
|
| 904 |
+
# This tells the estimator to run through the entire set.
|
| 905 |
+
eval_steps = None
|
| 906 |
+
# However, if running eval on the TPU, you will need to specify the
|
| 907 |
+
# number of steps.
|
| 908 |
+
if FLAGS.use_tpu:
|
| 909 |
+
assert len(eval_examples) % FLAGS.eval_batch_size == 0
|
| 910 |
+
eval_steps = int(len(eval_examples) // FLAGS.eval_batch_size)
|
| 911 |
+
|
| 912 |
+
eval_drop_remainder = True if FLAGS.use_tpu else False
|
| 913 |
+
eval_input_fn = file_based_input_fn_builder(
|
| 914 |
+
input_file=eval_file,
|
| 915 |
+
seq_length=FLAGS.max_seq_length,
|
| 916 |
+
is_training=False,
|
| 917 |
+
drop_remainder=eval_drop_remainder)
|
| 918 |
+
|
| 919 |
+
result = estimator.evaluate(input_fn=eval_input_fn, steps=eval_steps)
|
| 920 |
+
|
| 921 |
+
output_eval_file = os.path.join(FLAGS.output_dir, "eval_results.txt")
|
| 922 |
+
with tf.gfile.GFile(output_eval_file, "w") as writer:
|
| 923 |
+
tf.logging.info("***** Eval results *****")
|
| 924 |
+
for key in sorted(result.keys()):
|
| 925 |
+
tf.logging.info(" %s = %s", key, str(result[key]))
|
| 926 |
+
writer.write("%s = %s\n" % (key, str(result[key])))
|
| 927 |
+
|
| 928 |
+
if FLAGS.do_predict:
|
| 929 |
+
predict_examples = processor.get_test_examples(FLAGS.data_dir)
|
| 930 |
+
num_actual_predict_examples = len(predict_examples)
|
| 931 |
+
if FLAGS.use_tpu:
|
| 932 |
+
# TPU requires a fixed batch size for all batches, therefore the number
|
| 933 |
+
# of examples must be a multiple of the batch size, or else examples
|
| 934 |
+
# will get dropped. So we pad with fake examples which are ignored
|
| 935 |
+
# later on.
|
| 936 |
+
while len(predict_examples) % FLAGS.predict_batch_size != 0:
|
| 937 |
+
predict_examples.append(PaddingInputExample())
|
| 938 |
+
|
| 939 |
+
predict_file = os.path.join(FLAGS.output_dir, "predict.tf_record")
|
| 940 |
+
file_based_convert_examples_to_features(predict_examples, label_list,
|
| 941 |
+
FLAGS.max_seq_length, tokenizer,
|
| 942 |
+
predict_file)
|
| 943 |
+
|
| 944 |
+
tf.logging.info("***** Running prediction*****")
|
| 945 |
+
tf.logging.info(" Num examples = %d (%d actual, %d padding)",
|
| 946 |
+
len(predict_examples), num_actual_predict_examples,
|
| 947 |
+
len(predict_examples) - num_actual_predict_examples)
|
| 948 |
+
tf.logging.info(" Batch size = %d", FLAGS.predict_batch_size)
|
| 949 |
+
|
| 950 |
+
predict_drop_remainder = True if FLAGS.use_tpu else False
|
| 951 |
+
predict_input_fn = file_based_input_fn_builder(
|
| 952 |
+
input_file=predict_file,
|
| 953 |
+
seq_length=FLAGS.max_seq_length,
|
| 954 |
+
is_training=False,
|
| 955 |
+
drop_remainder=predict_drop_remainder)
|
| 956 |
+
|
| 957 |
+
result = estimator.predict(input_fn=predict_input_fn)
|
| 958 |
+
|
| 959 |
+
output_predict_file = os.path.join(FLAGS.output_dir, "test_results.tsv")
|
| 960 |
+
with tf.gfile.GFile(output_predict_file, "w") as writer:
|
| 961 |
+
num_written_lines = 0
|
| 962 |
+
tf.logging.info("***** Predict results *****")
|
| 963 |
+
for (i, prediction) in enumerate(result):
|
| 964 |
+
probabilities = prediction["probabilities"]
|
| 965 |
+
if i >= num_actual_predict_examples:
|
| 966 |
+
break
|
| 967 |
+
output_line = "\t".join(
|
| 968 |
+
str(class_probability)
|
| 969 |
+
for class_probability in probabilities) + "\n"
|
| 970 |
+
writer.write(output_line)
|
| 971 |
+
num_written_lines += 1
|
| 972 |
+
assert num_written_lines == num_actual_predict_examples
|
| 973 |
+
|
| 974 |
+
|
| 975 |
+
if __name__ == "__main__":
|
| 976 |
+
flags.mark_flag_as_required("data_dir")
|
| 977 |
+
flags.mark_flag_as_required("task_name")
|
| 978 |
+
flags.mark_flag_as_required("vocab_file")
|
| 979 |
+
flags.mark_flag_as_required("bert_config_file")
|
| 980 |
+
flags.mark_flag_as_required("output_dir")
|
| 981 |
+
tf.app.run()
|
bert-master/bert-master/run_classifier_with_tfhub.py
ADDED
|
@@ -0,0 +1,314 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# coding=utf-8
|
| 2 |
+
# Copyright 2018 The Google AI Language Team Authors.
|
| 3 |
+
#
|
| 4 |
+
# Licensed under the Apache License, Version 2.0 (the "License");
|
| 5 |
+
# you may not use this file except in compliance with the License.
|
| 6 |
+
# You may obtain a copy of the License at
|
| 7 |
+
#
|
| 8 |
+
# http://www.apache.org/licenses/LICENSE-2.0
|
| 9 |
+
#
|
| 10 |
+
# Unless required by applicable law or agreed to in writing, software
|
| 11 |
+
# distributed under the License is distributed on an "AS IS" BASIS,
|
| 12 |
+
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
| 13 |
+
# See the License for the specific language governing permissions and
|
| 14 |
+
# limitations under the License.
|
| 15 |
+
"""BERT finetuning runner with TF-Hub."""
|
| 16 |
+
|
| 17 |
+
from __future__ import absolute_import
|
| 18 |
+
from __future__ import division
|
| 19 |
+
from __future__ import print_function
|
| 20 |
+
|
| 21 |
+
import os
|
| 22 |
+
import optimization
|
| 23 |
+
import run_classifier
|
| 24 |
+
import tokenization
|
| 25 |
+
import tensorflow as tf
|
| 26 |
+
import tensorflow_hub as hub
|
| 27 |
+
|
| 28 |
+
flags = tf.flags
|
| 29 |
+
|
| 30 |
+
FLAGS = flags.FLAGS
|
| 31 |
+
|
| 32 |
+
flags.DEFINE_string(
|
| 33 |
+
"bert_hub_module_handle", None,
|
| 34 |
+
"Handle for the BERT TF-Hub module.")
|
| 35 |
+
|
| 36 |
+
|
| 37 |
+
def create_model(is_training, input_ids, input_mask, segment_ids, labels,
|
| 38 |
+
num_labels, bert_hub_module_handle):
|
| 39 |
+
"""Creates a classification model."""
|
| 40 |
+
tags = set()
|
| 41 |
+
if is_training:
|
| 42 |
+
tags.add("train")
|
| 43 |
+
bert_module = hub.Module(bert_hub_module_handle, tags=tags, trainable=True)
|
| 44 |
+
bert_inputs = dict(
|
| 45 |
+
input_ids=input_ids,
|
| 46 |
+
input_mask=input_mask,
|
| 47 |
+
segment_ids=segment_ids)
|
| 48 |
+
bert_outputs = bert_module(
|
| 49 |
+
inputs=bert_inputs,
|
| 50 |
+
signature="tokens",
|
| 51 |
+
as_dict=True)
|
| 52 |
+
|
| 53 |
+
# In the demo, we are doing a simple classification task on the entire
|
| 54 |
+
# segment.
|
| 55 |
+
#
|
| 56 |
+
# If you want to use the token-level output, use
|
| 57 |
+
# bert_outputs["sequence_output"] instead.
|
| 58 |
+
output_layer = bert_outputs["pooled_output"]
|
| 59 |
+
|
| 60 |
+
hidden_size = output_layer.shape[-1].value
|
| 61 |
+
|
| 62 |
+
output_weights = tf.get_variable(
|
| 63 |
+
"output_weights", [num_labels, hidden_size],
|
| 64 |
+
initializer=tf.truncated_normal_initializer(stddev=0.02))
|
| 65 |
+
|
| 66 |
+
output_bias = tf.get_variable(
|
| 67 |
+
"output_bias", [num_labels], initializer=tf.zeros_initializer())
|
| 68 |
+
|
| 69 |
+
with tf.variable_scope("loss"):
|
| 70 |
+
if is_training:
|
| 71 |
+
# I.e., 0.1 dropout
|
| 72 |
+
output_layer = tf.nn.dropout(output_layer, keep_prob=0.9)
|
| 73 |
+
|
| 74 |
+
logits = tf.matmul(output_layer, output_weights, transpose_b=True)
|
| 75 |
+
logits = tf.nn.bias_add(logits, output_bias)
|
| 76 |
+
probabilities = tf.nn.softmax(logits, axis=-1)
|
| 77 |
+
log_probs = tf.nn.log_softmax(logits, axis=-1)
|
| 78 |
+
|
| 79 |
+
one_hot_labels = tf.one_hot(labels, depth=num_labels, dtype=tf.float32)
|
| 80 |
+
|
| 81 |
+
per_example_loss = -tf.reduce_sum(one_hot_labels * log_probs, axis=-1)
|
| 82 |
+
loss = tf.reduce_mean(per_example_loss)
|
| 83 |
+
|
| 84 |
+
return (loss, per_example_loss, logits, probabilities)
|
| 85 |
+
|
| 86 |
+
|
| 87 |
+
def model_fn_builder(num_labels, learning_rate, num_train_steps,
|
| 88 |
+
num_warmup_steps, use_tpu, bert_hub_module_handle):
|
| 89 |
+
"""Returns `model_fn` closure for TPUEstimator."""
|
| 90 |
+
|
| 91 |
+
def model_fn(features, labels, mode, params): # pylint: disable=unused-argument
|
| 92 |
+
"""The `model_fn` for TPUEstimator."""
|
| 93 |
+
|
| 94 |
+
tf.logging.info("*** Features ***")
|
| 95 |
+
for name in sorted(features.keys()):
|
| 96 |
+
tf.logging.info(" name = %s, shape = %s" % (name, features[name].shape))
|
| 97 |
+
|
| 98 |
+
input_ids = features["input_ids"]
|
| 99 |
+
input_mask = features["input_mask"]
|
| 100 |
+
segment_ids = features["segment_ids"]
|
| 101 |
+
label_ids = features["label_ids"]
|
| 102 |
+
|
| 103 |
+
is_training = (mode == tf.estimator.ModeKeys.TRAIN)
|
| 104 |
+
|
| 105 |
+
(total_loss, per_example_loss, logits, probabilities) = create_model(
|
| 106 |
+
is_training, input_ids, input_mask, segment_ids, label_ids, num_labels,
|
| 107 |
+
bert_hub_module_handle)
|
| 108 |
+
|
| 109 |
+
output_spec = None
|
| 110 |
+
if mode == tf.estimator.ModeKeys.TRAIN:
|
| 111 |
+
train_op = optimization.create_optimizer(
|
| 112 |
+
total_loss, learning_rate, num_train_steps, num_warmup_steps, use_tpu)
|
| 113 |
+
|
| 114 |
+
output_spec = tf.contrib.tpu.TPUEstimatorSpec(
|
| 115 |
+
mode=mode,
|
| 116 |
+
loss=total_loss,
|
| 117 |
+
train_op=train_op)
|
| 118 |
+
elif mode == tf.estimator.ModeKeys.EVAL:
|
| 119 |
+
|
| 120 |
+
def metric_fn(per_example_loss, label_ids, logits):
|
| 121 |
+
predictions = tf.argmax(logits, axis=-1, output_type=tf.int32)
|
| 122 |
+
accuracy = tf.metrics.accuracy(label_ids, predictions)
|
| 123 |
+
loss = tf.metrics.mean(per_example_loss)
|
| 124 |
+
return {
|
| 125 |
+
"eval_accuracy": accuracy,
|
| 126 |
+
"eval_loss": loss,
|
| 127 |
+
}
|
| 128 |
+
|
| 129 |
+
eval_metrics = (metric_fn, [per_example_loss, label_ids, logits])
|
| 130 |
+
output_spec = tf.contrib.tpu.TPUEstimatorSpec(
|
| 131 |
+
mode=mode,
|
| 132 |
+
loss=total_loss,
|
| 133 |
+
eval_metrics=eval_metrics)
|
| 134 |
+
elif mode == tf.estimator.ModeKeys.PREDICT:
|
| 135 |
+
output_spec = tf.contrib.tpu.TPUEstimatorSpec(
|
| 136 |
+
mode=mode, predictions={"probabilities": probabilities})
|
| 137 |
+
else:
|
| 138 |
+
raise ValueError(
|
| 139 |
+
"Only TRAIN, EVAL and PREDICT modes are supported: %s" % (mode))
|
| 140 |
+
|
| 141 |
+
return output_spec
|
| 142 |
+
|
| 143 |
+
return model_fn
|
| 144 |
+
|
| 145 |
+
|
| 146 |
+
def create_tokenizer_from_hub_module(bert_hub_module_handle):
|
| 147 |
+
"""Get the vocab file and casing info from the Hub module."""
|
| 148 |
+
with tf.Graph().as_default():
|
| 149 |
+
bert_module = hub.Module(bert_hub_module_handle)
|
| 150 |
+
tokenization_info = bert_module(signature="tokenization_info", as_dict=True)
|
| 151 |
+
with tf.Session() as sess:
|
| 152 |
+
vocab_file, do_lower_case = sess.run([tokenization_info["vocab_file"],
|
| 153 |
+
tokenization_info["do_lower_case"]])
|
| 154 |
+
return tokenization.FullTokenizer(
|
| 155 |
+
vocab_file=vocab_file, do_lower_case=do_lower_case)
|
| 156 |
+
|
| 157 |
+
|
| 158 |
+
def main(_):
|
| 159 |
+
tf.logging.set_verbosity(tf.logging.INFO)
|
| 160 |
+
|
| 161 |
+
processors = {
|
| 162 |
+
"cola": run_classifier.ColaProcessor,
|
| 163 |
+
"mnli": run_classifier.MnliProcessor,
|
| 164 |
+
"mrpc": run_classifier.MrpcProcessor,
|
| 165 |
+
}
|
| 166 |
+
|
| 167 |
+
if not FLAGS.do_train and not FLAGS.do_eval:
|
| 168 |
+
raise ValueError("At least one of `do_train` or `do_eval` must be True.")
|
| 169 |
+
|
| 170 |
+
tf.gfile.MakeDirs(FLAGS.output_dir)
|
| 171 |
+
|
| 172 |
+
task_name = FLAGS.task_name.lower()
|
| 173 |
+
|
| 174 |
+
if task_name not in processors:
|
| 175 |
+
raise ValueError("Task not found: %s" % (task_name))
|
| 176 |
+
|
| 177 |
+
processor = processors[task_name]()
|
| 178 |
+
|
| 179 |
+
label_list = processor.get_labels()
|
| 180 |
+
|
| 181 |
+
tokenizer = create_tokenizer_from_hub_module(FLAGS.bert_hub_module_handle)
|
| 182 |
+
|
| 183 |
+
tpu_cluster_resolver = None
|
| 184 |
+
if FLAGS.use_tpu and FLAGS.tpu_name:
|
| 185 |
+
tpu_cluster_resolver = tf.contrib.cluster_resolver.TPUClusterResolver(
|
| 186 |
+
FLAGS.tpu_name, zone=FLAGS.tpu_zone, project=FLAGS.gcp_project)
|
| 187 |
+
|
| 188 |
+
is_per_host = tf.contrib.tpu.InputPipelineConfig.PER_HOST_V2
|
| 189 |
+
run_config = tf.contrib.tpu.RunConfig(
|
| 190 |
+
cluster=tpu_cluster_resolver,
|
| 191 |
+
master=FLAGS.master,
|
| 192 |
+
model_dir=FLAGS.output_dir,
|
| 193 |
+
save_checkpoints_steps=FLAGS.save_checkpoints_steps,
|
| 194 |
+
tpu_config=tf.contrib.tpu.TPUConfig(
|
| 195 |
+
iterations_per_loop=FLAGS.iterations_per_loop,
|
| 196 |
+
num_shards=FLAGS.num_tpu_cores,
|
| 197 |
+
per_host_input_for_training=is_per_host))
|
| 198 |
+
|
| 199 |
+
train_examples = None
|
| 200 |
+
num_train_steps = None
|
| 201 |
+
num_warmup_steps = None
|
| 202 |
+
if FLAGS.do_train:
|
| 203 |
+
train_examples = processor.get_train_examples(FLAGS.data_dir)
|
| 204 |
+
num_train_steps = int(
|
| 205 |
+
len(train_examples) / FLAGS.train_batch_size * FLAGS.num_train_epochs)
|
| 206 |
+
num_warmup_steps = int(num_train_steps * FLAGS.warmup_proportion)
|
| 207 |
+
|
| 208 |
+
model_fn = model_fn_builder(
|
| 209 |
+
num_labels=len(label_list),
|
| 210 |
+
learning_rate=FLAGS.learning_rate,
|
| 211 |
+
num_train_steps=num_train_steps,
|
| 212 |
+
num_warmup_steps=num_warmup_steps,
|
| 213 |
+
use_tpu=FLAGS.use_tpu,
|
| 214 |
+
bert_hub_module_handle=FLAGS.bert_hub_module_handle)
|
| 215 |
+
|
| 216 |
+
# If TPU is not available, this will fall back to normal Estimator on CPU
|
| 217 |
+
# or GPU.
|
| 218 |
+
estimator = tf.contrib.tpu.TPUEstimator(
|
| 219 |
+
use_tpu=FLAGS.use_tpu,
|
| 220 |
+
model_fn=model_fn,
|
| 221 |
+
config=run_config,
|
| 222 |
+
train_batch_size=FLAGS.train_batch_size,
|
| 223 |
+
eval_batch_size=FLAGS.eval_batch_size,
|
| 224 |
+
predict_batch_size=FLAGS.predict_batch_size)
|
| 225 |
+
|
| 226 |
+
if FLAGS.do_train:
|
| 227 |
+
train_features = run_classifier.convert_examples_to_features(
|
| 228 |
+
train_examples, label_list, FLAGS.max_seq_length, tokenizer)
|
| 229 |
+
tf.logging.info("***** Running training *****")
|
| 230 |
+
tf.logging.info(" Num examples = %d", len(train_examples))
|
| 231 |
+
tf.logging.info(" Batch size = %d", FLAGS.train_batch_size)
|
| 232 |
+
tf.logging.info(" Num steps = %d", num_train_steps)
|
| 233 |
+
train_input_fn = run_classifier.input_fn_builder(
|
| 234 |
+
features=train_features,
|
| 235 |
+
seq_length=FLAGS.max_seq_length,
|
| 236 |
+
is_training=True,
|
| 237 |
+
drop_remainder=True)
|
| 238 |
+
estimator.train(input_fn=train_input_fn, max_steps=num_train_steps)
|
| 239 |
+
|
| 240 |
+
if FLAGS.do_eval:
|
| 241 |
+
eval_examples = processor.get_dev_examples(FLAGS.data_dir)
|
| 242 |
+
eval_features = run_classifier.convert_examples_to_features(
|
| 243 |
+
eval_examples, label_list, FLAGS.max_seq_length, tokenizer)
|
| 244 |
+
|
| 245 |
+
tf.logging.info("***** Running evaluation *****")
|
| 246 |
+
tf.logging.info(" Num examples = %d", len(eval_examples))
|
| 247 |
+
tf.logging.info(" Batch size = %d", FLAGS.eval_batch_size)
|
| 248 |
+
|
| 249 |
+
# This tells the estimator to run through the entire set.
|
| 250 |
+
eval_steps = None
|
| 251 |
+
# However, if running eval on the TPU, you will need to specify the
|
| 252 |
+
# number of steps.
|
| 253 |
+
if FLAGS.use_tpu:
|
| 254 |
+
# Eval will be slightly WRONG on the TPU because it will truncate
|
| 255 |
+
# the last batch.
|
| 256 |
+
eval_steps = int(len(eval_examples) / FLAGS.eval_batch_size)
|
| 257 |
+
|
| 258 |
+
eval_drop_remainder = True if FLAGS.use_tpu else False
|
| 259 |
+
eval_input_fn = run_classifier.input_fn_builder(
|
| 260 |
+
features=eval_features,
|
| 261 |
+
seq_length=FLAGS.max_seq_length,
|
| 262 |
+
is_training=False,
|
| 263 |
+
drop_remainder=eval_drop_remainder)
|
| 264 |
+
|
| 265 |
+
result = estimator.evaluate(input_fn=eval_input_fn, steps=eval_steps)
|
| 266 |
+
|
| 267 |
+
output_eval_file = os.path.join(FLAGS.output_dir, "eval_results.txt")
|
| 268 |
+
with tf.gfile.GFile(output_eval_file, "w") as writer:
|
| 269 |
+
tf.logging.info("***** Eval results *****")
|
| 270 |
+
for key in sorted(result.keys()):
|
| 271 |
+
tf.logging.info(" %s = %s", key, str(result[key]))
|
| 272 |
+
writer.write("%s = %s\n" % (key, str(result[key])))
|
| 273 |
+
|
| 274 |
+
if FLAGS.do_predict:
|
| 275 |
+
predict_examples = processor.get_test_examples(FLAGS.data_dir)
|
| 276 |
+
if FLAGS.use_tpu:
|
| 277 |
+
# Discard batch remainder if running on TPU
|
| 278 |
+
n = len(predict_examples)
|
| 279 |
+
predict_examples = predict_examples[:(n - n % FLAGS.predict_batch_size)]
|
| 280 |
+
|
| 281 |
+
predict_file = os.path.join(FLAGS.output_dir, "predict.tf_record")
|
| 282 |
+
run_classifier.file_based_convert_examples_to_features(
|
| 283 |
+
predict_examples, label_list, FLAGS.max_seq_length, tokenizer,
|
| 284 |
+
predict_file)
|
| 285 |
+
|
| 286 |
+
tf.logging.info("***** Running prediction*****")
|
| 287 |
+
tf.logging.info(" Num examples = %d", len(predict_examples))
|
| 288 |
+
tf.logging.info(" Batch size = %d", FLAGS.predict_batch_size)
|
| 289 |
+
|
| 290 |
+
predict_input_fn = run_classifier.file_based_input_fn_builder(
|
| 291 |
+
input_file=predict_file,
|
| 292 |
+
seq_length=FLAGS.max_seq_length,
|
| 293 |
+
is_training=False,
|
| 294 |
+
drop_remainder=FLAGS.use_tpu)
|
| 295 |
+
|
| 296 |
+
result = estimator.predict(input_fn=predict_input_fn)
|
| 297 |
+
|
| 298 |
+
output_predict_file = os.path.join(FLAGS.output_dir, "test_results.tsv")
|
| 299 |
+
with tf.gfile.GFile(output_predict_file, "w") as writer:
|
| 300 |
+
tf.logging.info("***** Predict results *****")
|
| 301 |
+
for prediction in result:
|
| 302 |
+
probabilities = prediction["probabilities"]
|
| 303 |
+
output_line = "\t".join(
|
| 304 |
+
str(class_probability)
|
| 305 |
+
for class_probability in probabilities) + "\n"
|
| 306 |
+
writer.write(output_line)
|
| 307 |
+
|
| 308 |
+
|
| 309 |
+
if __name__ == "__main__":
|
| 310 |
+
flags.mark_flag_as_required("data_dir")
|
| 311 |
+
flags.mark_flag_as_required("task_name")
|
| 312 |
+
flags.mark_flag_as_required("bert_hub_module_handle")
|
| 313 |
+
flags.mark_flag_as_required("output_dir")
|
| 314 |
+
tf.app.run()
|
bert-master/bert-master/run_pretraining.py
ADDED
|
@@ -0,0 +1,493 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# coding=utf-8
|
| 2 |
+
# Copyright 2018 The Google AI Language Team Authors.
|
| 3 |
+
#
|
| 4 |
+
# Licensed under the Apache License, Version 2.0 (the "License");
|
| 5 |
+
# you may not use this file except in compliance with the License.
|
| 6 |
+
# You may obtain a copy of the License at
|
| 7 |
+
#
|
| 8 |
+
# http://www.apache.org/licenses/LICENSE-2.0
|
| 9 |
+
#
|
| 10 |
+
# Unless required by applicable law or agreed to in writing, software
|
| 11 |
+
# distributed under the License is distributed on an "AS IS" BASIS,
|
| 12 |
+
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
| 13 |
+
# See the License for the specific language governing permissions and
|
| 14 |
+
# limitations under the License.
|
| 15 |
+
"""Run masked LM/next sentence masked_lm pre-training for BERT."""
|
| 16 |
+
|
| 17 |
+
from __future__ import absolute_import
|
| 18 |
+
from __future__ import division
|
| 19 |
+
from __future__ import print_function
|
| 20 |
+
|
| 21 |
+
import os
|
| 22 |
+
import modeling
|
| 23 |
+
import optimization
|
| 24 |
+
import tensorflow as tf
|
| 25 |
+
|
| 26 |
+
flags = tf.flags
|
| 27 |
+
|
| 28 |
+
FLAGS = flags.FLAGS
|
| 29 |
+
|
| 30 |
+
## Required parameters
|
| 31 |
+
flags.DEFINE_string(
|
| 32 |
+
"bert_config_file", None,
|
| 33 |
+
"The config json file corresponding to the pre-trained BERT model. "
|
| 34 |
+
"This specifies the model architecture.")
|
| 35 |
+
|
| 36 |
+
flags.DEFINE_string(
|
| 37 |
+
"input_file", None,
|
| 38 |
+
"Input TF example files (can be a glob or comma separated).")
|
| 39 |
+
|
| 40 |
+
flags.DEFINE_string(
|
| 41 |
+
"output_dir", None,
|
| 42 |
+
"The output directory where the model checkpoints will be written.")
|
| 43 |
+
|
| 44 |
+
## Other parameters
|
| 45 |
+
flags.DEFINE_string(
|
| 46 |
+
"init_checkpoint", None,
|
| 47 |
+
"Initial checkpoint (usually from a pre-trained BERT model).")
|
| 48 |
+
|
| 49 |
+
flags.DEFINE_integer(
|
| 50 |
+
"max_seq_length", 128,
|
| 51 |
+
"The maximum total input sequence length after WordPiece tokenization. "
|
| 52 |
+
"Sequences longer than this will be truncated, and sequences shorter "
|
| 53 |
+
"than this will be padded. Must match data generation.")
|
| 54 |
+
|
| 55 |
+
flags.DEFINE_integer(
|
| 56 |
+
"max_predictions_per_seq", 20,
|
| 57 |
+
"Maximum number of masked LM predictions per sequence. "
|
| 58 |
+
"Must match data generation.")
|
| 59 |
+
|
| 60 |
+
flags.DEFINE_bool("do_train", False, "Whether to run training.")
|
| 61 |
+
|
| 62 |
+
flags.DEFINE_bool("do_eval", False, "Whether to run eval on the dev set.")
|
| 63 |
+
|
| 64 |
+
flags.DEFINE_integer("train_batch_size", 32, "Total batch size for training.")
|
| 65 |
+
|
| 66 |
+
flags.DEFINE_integer("eval_batch_size", 8, "Total batch size for eval.")
|
| 67 |
+
|
| 68 |
+
flags.DEFINE_float("learning_rate", 5e-5, "The initial learning rate for Adam.")
|
| 69 |
+
|
| 70 |
+
flags.DEFINE_integer("num_train_steps", 100000, "Number of training steps.")
|
| 71 |
+
|
| 72 |
+
flags.DEFINE_integer("num_warmup_steps", 10000, "Number of warmup steps.")
|
| 73 |
+
|
| 74 |
+
flags.DEFINE_integer("save_checkpoints_steps", 1000,
|
| 75 |
+
"How often to save the model checkpoint.")
|
| 76 |
+
|
| 77 |
+
flags.DEFINE_integer("iterations_per_loop", 1000,
|
| 78 |
+
"How many steps to make in each estimator call.")
|
| 79 |
+
|
| 80 |
+
flags.DEFINE_integer("max_eval_steps", 100, "Maximum number of eval steps.")
|
| 81 |
+
|
| 82 |
+
flags.DEFINE_bool("use_tpu", False, "Whether to use TPU or GPU/CPU.")
|
| 83 |
+
|
| 84 |
+
tf.flags.DEFINE_string(
|
| 85 |
+
"tpu_name", None,
|
| 86 |
+
"The Cloud TPU to use for training. This should be either the name "
|
| 87 |
+
"used when creating the Cloud TPU, or a grpc://ip.address.of.tpu:8470 "
|
| 88 |
+
"url.")
|
| 89 |
+
|
| 90 |
+
tf.flags.DEFINE_string(
|
| 91 |
+
"tpu_zone", None,
|
| 92 |
+
"[Optional] GCE zone where the Cloud TPU is located in. If not "
|
| 93 |
+
"specified, we will attempt to automatically detect the GCE project from "
|
| 94 |
+
"metadata.")
|
| 95 |
+
|
| 96 |
+
tf.flags.DEFINE_string(
|
| 97 |
+
"gcp_project", None,
|
| 98 |
+
"[Optional] Project name for the Cloud TPU-enabled project. If not "
|
| 99 |
+
"specified, we will attempt to automatically detect the GCE project from "
|
| 100 |
+
"metadata.")
|
| 101 |
+
|
| 102 |
+
tf.flags.DEFINE_string("master", None, "[Optional] TensorFlow master URL.")
|
| 103 |
+
|
| 104 |
+
flags.DEFINE_integer(
|
| 105 |
+
"num_tpu_cores", 8,
|
| 106 |
+
"Only used if `use_tpu` is True. Total number of TPU cores to use.")
|
| 107 |
+
|
| 108 |
+
|
| 109 |
+
def model_fn_builder(bert_config, init_checkpoint, learning_rate,
|
| 110 |
+
num_train_steps, num_warmup_steps, use_tpu,
|
| 111 |
+
use_one_hot_embeddings):
|
| 112 |
+
"""Returns `model_fn` closure for TPUEstimator."""
|
| 113 |
+
|
| 114 |
+
def model_fn(features, labels, mode, params): # pylint: disable=unused-argument
|
| 115 |
+
"""The `model_fn` for TPUEstimator."""
|
| 116 |
+
|
| 117 |
+
tf.logging.info("*** Features ***")
|
| 118 |
+
for name in sorted(features.keys()):
|
| 119 |
+
tf.logging.info(" name = %s, shape = %s" % (name, features[name].shape))
|
| 120 |
+
|
| 121 |
+
input_ids = features["input_ids"]
|
| 122 |
+
input_mask = features["input_mask"]
|
| 123 |
+
segment_ids = features["segment_ids"]
|
| 124 |
+
masked_lm_positions = features["masked_lm_positions"]
|
| 125 |
+
masked_lm_ids = features["masked_lm_ids"]
|
| 126 |
+
masked_lm_weights = features["masked_lm_weights"]
|
| 127 |
+
next_sentence_labels = features["next_sentence_labels"]
|
| 128 |
+
|
| 129 |
+
is_training = (mode == tf.estimator.ModeKeys.TRAIN)
|
| 130 |
+
|
| 131 |
+
model = modeling.BertModel(
|
| 132 |
+
config=bert_config,
|
| 133 |
+
is_training=is_training,
|
| 134 |
+
input_ids=input_ids,
|
| 135 |
+
input_mask=input_mask,
|
| 136 |
+
token_type_ids=segment_ids,
|
| 137 |
+
use_one_hot_embeddings=use_one_hot_embeddings)
|
| 138 |
+
|
| 139 |
+
(masked_lm_loss,
|
| 140 |
+
masked_lm_example_loss, masked_lm_log_probs) = get_masked_lm_output(
|
| 141 |
+
bert_config, model.get_sequence_output(), model.get_embedding_table(),
|
| 142 |
+
masked_lm_positions, masked_lm_ids, masked_lm_weights)
|
| 143 |
+
|
| 144 |
+
(next_sentence_loss, next_sentence_example_loss,
|
| 145 |
+
next_sentence_log_probs) = get_next_sentence_output(
|
| 146 |
+
bert_config, model.get_pooled_output(), next_sentence_labels)
|
| 147 |
+
|
| 148 |
+
total_loss = masked_lm_loss + next_sentence_loss
|
| 149 |
+
|
| 150 |
+
tvars = tf.trainable_variables()
|
| 151 |
+
|
| 152 |
+
initialized_variable_names = {}
|
| 153 |
+
scaffold_fn = None
|
| 154 |
+
if init_checkpoint:
|
| 155 |
+
(assignment_map, initialized_variable_names
|
| 156 |
+
) = modeling.get_assignment_map_from_checkpoint(tvars, init_checkpoint)
|
| 157 |
+
if use_tpu:
|
| 158 |
+
|
| 159 |
+
def tpu_scaffold():
|
| 160 |
+
tf.train.init_from_checkpoint(init_checkpoint, assignment_map)
|
| 161 |
+
return tf.train.Scaffold()
|
| 162 |
+
|
| 163 |
+
scaffold_fn = tpu_scaffold
|
| 164 |
+
else:
|
| 165 |
+
tf.train.init_from_checkpoint(init_checkpoint, assignment_map)
|
| 166 |
+
|
| 167 |
+
tf.logging.info("**** Trainable Variables ****")
|
| 168 |
+
for var in tvars:
|
| 169 |
+
init_string = ""
|
| 170 |
+
if var.name in initialized_variable_names:
|
| 171 |
+
init_string = ", *INIT_FROM_CKPT*"
|
| 172 |
+
tf.logging.info(" name = %s, shape = %s%s", var.name, var.shape,
|
| 173 |
+
init_string)
|
| 174 |
+
|
| 175 |
+
output_spec = None
|
| 176 |
+
if mode == tf.estimator.ModeKeys.TRAIN:
|
| 177 |
+
train_op = optimization.create_optimizer(
|
| 178 |
+
total_loss, learning_rate, num_train_steps, num_warmup_steps, use_tpu)
|
| 179 |
+
|
| 180 |
+
output_spec = tf.contrib.tpu.TPUEstimatorSpec(
|
| 181 |
+
mode=mode,
|
| 182 |
+
loss=total_loss,
|
| 183 |
+
train_op=train_op,
|
| 184 |
+
scaffold_fn=scaffold_fn)
|
| 185 |
+
elif mode == tf.estimator.ModeKeys.EVAL:
|
| 186 |
+
|
| 187 |
+
def metric_fn(masked_lm_example_loss, masked_lm_log_probs, masked_lm_ids,
|
| 188 |
+
masked_lm_weights, next_sentence_example_loss,
|
| 189 |
+
next_sentence_log_probs, next_sentence_labels):
|
| 190 |
+
"""Computes the loss and accuracy of the model."""
|
| 191 |
+
masked_lm_log_probs = tf.reshape(masked_lm_log_probs,
|
| 192 |
+
[-1, masked_lm_log_probs.shape[-1]])
|
| 193 |
+
masked_lm_predictions = tf.argmax(
|
| 194 |
+
masked_lm_log_probs, axis=-1, output_type=tf.int32)
|
| 195 |
+
masked_lm_example_loss = tf.reshape(masked_lm_example_loss, [-1])
|
| 196 |
+
masked_lm_ids = tf.reshape(masked_lm_ids, [-1])
|
| 197 |
+
masked_lm_weights = tf.reshape(masked_lm_weights, [-1])
|
| 198 |
+
masked_lm_accuracy = tf.metrics.accuracy(
|
| 199 |
+
labels=masked_lm_ids,
|
| 200 |
+
predictions=masked_lm_predictions,
|
| 201 |
+
weights=masked_lm_weights)
|
| 202 |
+
masked_lm_mean_loss = tf.metrics.mean(
|
| 203 |
+
values=masked_lm_example_loss, weights=masked_lm_weights)
|
| 204 |
+
|
| 205 |
+
next_sentence_log_probs = tf.reshape(
|
| 206 |
+
next_sentence_log_probs, [-1, next_sentence_log_probs.shape[-1]])
|
| 207 |
+
next_sentence_predictions = tf.argmax(
|
| 208 |
+
next_sentence_log_probs, axis=-1, output_type=tf.int32)
|
| 209 |
+
next_sentence_labels = tf.reshape(next_sentence_labels, [-1])
|
| 210 |
+
next_sentence_accuracy = tf.metrics.accuracy(
|
| 211 |
+
labels=next_sentence_labels, predictions=next_sentence_predictions)
|
| 212 |
+
next_sentence_mean_loss = tf.metrics.mean(
|
| 213 |
+
values=next_sentence_example_loss)
|
| 214 |
+
|
| 215 |
+
return {
|
| 216 |
+
"masked_lm_accuracy": masked_lm_accuracy,
|
| 217 |
+
"masked_lm_loss": masked_lm_mean_loss,
|
| 218 |
+
"next_sentence_accuracy": next_sentence_accuracy,
|
| 219 |
+
"next_sentence_loss": next_sentence_mean_loss,
|
| 220 |
+
}
|
| 221 |
+
|
| 222 |
+
eval_metrics = (metric_fn, [
|
| 223 |
+
masked_lm_example_loss, masked_lm_log_probs, masked_lm_ids,
|
| 224 |
+
masked_lm_weights, next_sentence_example_loss,
|
| 225 |
+
next_sentence_log_probs, next_sentence_labels
|
| 226 |
+
])
|
| 227 |
+
output_spec = tf.contrib.tpu.TPUEstimatorSpec(
|
| 228 |
+
mode=mode,
|
| 229 |
+
loss=total_loss,
|
| 230 |
+
eval_metrics=eval_metrics,
|
| 231 |
+
scaffold_fn=scaffold_fn)
|
| 232 |
+
else:
|
| 233 |
+
raise ValueError("Only TRAIN and EVAL modes are supported: %s" % (mode))
|
| 234 |
+
|
| 235 |
+
return output_spec
|
| 236 |
+
|
| 237 |
+
return model_fn
|
| 238 |
+
|
| 239 |
+
|
| 240 |
+
def get_masked_lm_output(bert_config, input_tensor, output_weights, positions,
|
| 241 |
+
label_ids, label_weights):
|
| 242 |
+
"""Get loss and log probs for the masked LM."""
|
| 243 |
+
input_tensor = gather_indexes(input_tensor, positions)
|
| 244 |
+
|
| 245 |
+
with tf.variable_scope("cls/predictions"):
|
| 246 |
+
# We apply one more non-linear transformation before the output layer.
|
| 247 |
+
# This matrix is not used after pre-training.
|
| 248 |
+
with tf.variable_scope("transform"):
|
| 249 |
+
input_tensor = tf.layers.dense(
|
| 250 |
+
input_tensor,
|
| 251 |
+
units=bert_config.hidden_size,
|
| 252 |
+
activation=modeling.get_activation(bert_config.hidden_act),
|
| 253 |
+
kernel_initializer=modeling.create_initializer(
|
| 254 |
+
bert_config.initializer_range))
|
| 255 |
+
input_tensor = modeling.layer_norm(input_tensor)
|
| 256 |
+
|
| 257 |
+
# The output weights are the same as the input embeddings, but there is
|
| 258 |
+
# an output-only bias for each token.
|
| 259 |
+
output_bias = tf.get_variable(
|
| 260 |
+
"output_bias",
|
| 261 |
+
shape=[bert_config.vocab_size],
|
| 262 |
+
initializer=tf.zeros_initializer())
|
| 263 |
+
logits = tf.matmul(input_tensor, output_weights, transpose_b=True)
|
| 264 |
+
logits = tf.nn.bias_add(logits, output_bias)
|
| 265 |
+
log_probs = tf.nn.log_softmax(logits, axis=-1)
|
| 266 |
+
|
| 267 |
+
label_ids = tf.reshape(label_ids, [-1])
|
| 268 |
+
label_weights = tf.reshape(label_weights, [-1])
|
| 269 |
+
|
| 270 |
+
one_hot_labels = tf.one_hot(
|
| 271 |
+
label_ids, depth=bert_config.vocab_size, dtype=tf.float32)
|
| 272 |
+
|
| 273 |
+
# The `positions` tensor might be zero-padded (if the sequence is too
|
| 274 |
+
# short to have the maximum number of predictions). The `label_weights`
|
| 275 |
+
# tensor has a value of 1.0 for every real prediction and 0.0 for the
|
| 276 |
+
# padding predictions.
|
| 277 |
+
per_example_loss = -tf.reduce_sum(log_probs * one_hot_labels, axis=[-1])
|
| 278 |
+
numerator = tf.reduce_sum(label_weights * per_example_loss)
|
| 279 |
+
denominator = tf.reduce_sum(label_weights) + 1e-5
|
| 280 |
+
loss = numerator / denominator
|
| 281 |
+
|
| 282 |
+
return (loss, per_example_loss, log_probs)
|
| 283 |
+
|
| 284 |
+
|
| 285 |
+
def get_next_sentence_output(bert_config, input_tensor, labels):
|
| 286 |
+
"""Get loss and log probs for the next sentence prediction."""
|
| 287 |
+
|
| 288 |
+
# Simple binary classification. Note that 0 is "next sentence" and 1 is
|
| 289 |
+
# "random sentence". This weight matrix is not used after pre-training.
|
| 290 |
+
with tf.variable_scope("cls/seq_relationship"):
|
| 291 |
+
output_weights = tf.get_variable(
|
| 292 |
+
"output_weights",
|
| 293 |
+
shape=[2, bert_config.hidden_size],
|
| 294 |
+
initializer=modeling.create_initializer(bert_config.initializer_range))
|
| 295 |
+
output_bias = tf.get_variable(
|
| 296 |
+
"output_bias", shape=[2], initializer=tf.zeros_initializer())
|
| 297 |
+
|
| 298 |
+
logits = tf.matmul(input_tensor, output_weights, transpose_b=True)
|
| 299 |
+
logits = tf.nn.bias_add(logits, output_bias)
|
| 300 |
+
log_probs = tf.nn.log_softmax(logits, axis=-1)
|
| 301 |
+
labels = tf.reshape(labels, [-1])
|
| 302 |
+
one_hot_labels = tf.one_hot(labels, depth=2, dtype=tf.float32)
|
| 303 |
+
per_example_loss = -tf.reduce_sum(one_hot_labels * log_probs, axis=-1)
|
| 304 |
+
loss = tf.reduce_mean(per_example_loss)
|
| 305 |
+
return (loss, per_example_loss, log_probs)
|
| 306 |
+
|
| 307 |
+
|
| 308 |
+
def gather_indexes(sequence_tensor, positions):
|
| 309 |
+
"""Gathers the vectors at the specific positions over a minibatch."""
|
| 310 |
+
sequence_shape = modeling.get_shape_list(sequence_tensor, expected_rank=3)
|
| 311 |
+
batch_size = sequence_shape[0]
|
| 312 |
+
seq_length = sequence_shape[1]
|
| 313 |
+
width = sequence_shape[2]
|
| 314 |
+
|
| 315 |
+
flat_offsets = tf.reshape(
|
| 316 |
+
tf.range(0, batch_size, dtype=tf.int32) * seq_length, [-1, 1])
|
| 317 |
+
flat_positions = tf.reshape(positions + flat_offsets, [-1])
|
| 318 |
+
flat_sequence_tensor = tf.reshape(sequence_tensor,
|
| 319 |
+
[batch_size * seq_length, width])
|
| 320 |
+
output_tensor = tf.gather(flat_sequence_tensor, flat_positions)
|
| 321 |
+
return output_tensor
|
| 322 |
+
|
| 323 |
+
|
| 324 |
+
def input_fn_builder(input_files,
|
| 325 |
+
max_seq_length,
|
| 326 |
+
max_predictions_per_seq,
|
| 327 |
+
is_training,
|
| 328 |
+
num_cpu_threads=4):
|
| 329 |
+
"""Creates an `input_fn` closure to be passed to TPUEstimator."""
|
| 330 |
+
|
| 331 |
+
def input_fn(params):
|
| 332 |
+
"""The actual input function."""
|
| 333 |
+
batch_size = params["batch_size"]
|
| 334 |
+
|
| 335 |
+
name_to_features = {
|
| 336 |
+
"input_ids":
|
| 337 |
+
tf.FixedLenFeature([max_seq_length], tf.int64),
|
| 338 |
+
"input_mask":
|
| 339 |
+
tf.FixedLenFeature([max_seq_length], tf.int64),
|
| 340 |
+
"segment_ids":
|
| 341 |
+
tf.FixedLenFeature([max_seq_length], tf.int64),
|
| 342 |
+
"masked_lm_positions":
|
| 343 |
+
tf.FixedLenFeature([max_predictions_per_seq], tf.int64),
|
| 344 |
+
"masked_lm_ids":
|
| 345 |
+
tf.FixedLenFeature([max_predictions_per_seq], tf.int64),
|
| 346 |
+
"masked_lm_weights":
|
| 347 |
+
tf.FixedLenFeature([max_predictions_per_seq], tf.float32),
|
| 348 |
+
"next_sentence_labels":
|
| 349 |
+
tf.FixedLenFeature([1], tf.int64),
|
| 350 |
+
}
|
| 351 |
+
|
| 352 |
+
# For training, we want a lot of parallel reading and shuffling.
|
| 353 |
+
# For eval, we want no shuffling and parallel reading doesn't matter.
|
| 354 |
+
if is_training:
|
| 355 |
+
d = tf.data.Dataset.from_tensor_slices(tf.constant(input_files))
|
| 356 |
+
d = d.repeat()
|
| 357 |
+
d = d.shuffle(buffer_size=len(input_files))
|
| 358 |
+
|
| 359 |
+
# `cycle_length` is the number of parallel files that get read.
|
| 360 |
+
cycle_length = min(num_cpu_threads, len(input_files))
|
| 361 |
+
|
| 362 |
+
# `sloppy` mode means that the interleaving is not exact. This adds
|
| 363 |
+
# even more randomness to the training pipeline.
|
| 364 |
+
d = d.apply(
|
| 365 |
+
tf.contrib.data.parallel_interleave(
|
| 366 |
+
tf.data.TFRecordDataset,
|
| 367 |
+
sloppy=is_training,
|
| 368 |
+
cycle_length=cycle_length))
|
| 369 |
+
d = d.shuffle(buffer_size=100)
|
| 370 |
+
else:
|
| 371 |
+
d = tf.data.TFRecordDataset(input_files)
|
| 372 |
+
# Since we evaluate for a fixed number of steps we don't want to encounter
|
| 373 |
+
# out-of-range exceptions.
|
| 374 |
+
d = d.repeat()
|
| 375 |
+
|
| 376 |
+
# We must `drop_remainder` on training because the TPU requires fixed
|
| 377 |
+
# size dimensions. For eval, we assume we are evaluating on the CPU or GPU
|
| 378 |
+
# and we *don't* want to drop the remainder, otherwise we wont cover
|
| 379 |
+
# every sample.
|
| 380 |
+
d = d.apply(
|
| 381 |
+
tf.contrib.data.map_and_batch(
|
| 382 |
+
lambda record: _decode_record(record, name_to_features),
|
| 383 |
+
batch_size=batch_size,
|
| 384 |
+
num_parallel_batches=num_cpu_threads,
|
| 385 |
+
drop_remainder=True))
|
| 386 |
+
return d
|
| 387 |
+
|
| 388 |
+
return input_fn
|
| 389 |
+
|
| 390 |
+
|
| 391 |
+
def _decode_record(record, name_to_features):
|
| 392 |
+
"""Decodes a record to a TensorFlow example."""
|
| 393 |
+
example = tf.parse_single_example(record, name_to_features)
|
| 394 |
+
|
| 395 |
+
# tf.Example only supports tf.int64, but the TPU only supports tf.int32.
|
| 396 |
+
# So cast all int64 to int32.
|
| 397 |
+
for name in list(example.keys()):
|
| 398 |
+
t = example[name]
|
| 399 |
+
if t.dtype == tf.int64:
|
| 400 |
+
t = tf.to_int32(t)
|
| 401 |
+
example[name] = t
|
| 402 |
+
|
| 403 |
+
return example
|
| 404 |
+
|
| 405 |
+
|
| 406 |
+
def main(_):
|
| 407 |
+
tf.logging.set_verbosity(tf.logging.INFO)
|
| 408 |
+
|
| 409 |
+
if not FLAGS.do_train and not FLAGS.do_eval:
|
| 410 |
+
raise ValueError("At least one of `do_train` or `do_eval` must be True.")
|
| 411 |
+
|
| 412 |
+
bert_config = modeling.BertConfig.from_json_file(FLAGS.bert_config_file)
|
| 413 |
+
|
| 414 |
+
tf.gfile.MakeDirs(FLAGS.output_dir)
|
| 415 |
+
|
| 416 |
+
input_files = []
|
| 417 |
+
for input_pattern in FLAGS.input_file.split(","):
|
| 418 |
+
input_files.extend(tf.gfile.Glob(input_pattern))
|
| 419 |
+
|
| 420 |
+
tf.logging.info("*** Input Files ***")
|
| 421 |
+
for input_file in input_files:
|
| 422 |
+
tf.logging.info(" %s" % input_file)
|
| 423 |
+
|
| 424 |
+
tpu_cluster_resolver = None
|
| 425 |
+
if FLAGS.use_tpu and FLAGS.tpu_name:
|
| 426 |
+
tpu_cluster_resolver = tf.contrib.cluster_resolver.TPUClusterResolver(
|
| 427 |
+
FLAGS.tpu_name, zone=FLAGS.tpu_zone, project=FLAGS.gcp_project)
|
| 428 |
+
|
| 429 |
+
is_per_host = tf.contrib.tpu.InputPipelineConfig.PER_HOST_V2
|
| 430 |
+
run_config = tf.contrib.tpu.RunConfig(
|
| 431 |
+
cluster=tpu_cluster_resolver,
|
| 432 |
+
master=FLAGS.master,
|
| 433 |
+
model_dir=FLAGS.output_dir,
|
| 434 |
+
save_checkpoints_steps=FLAGS.save_checkpoints_steps,
|
| 435 |
+
tpu_config=tf.contrib.tpu.TPUConfig(
|
| 436 |
+
iterations_per_loop=FLAGS.iterations_per_loop,
|
| 437 |
+
num_shards=FLAGS.num_tpu_cores,
|
| 438 |
+
per_host_input_for_training=is_per_host))
|
| 439 |
+
|
| 440 |
+
model_fn = model_fn_builder(
|
| 441 |
+
bert_config=bert_config,
|
| 442 |
+
init_checkpoint=FLAGS.init_checkpoint,
|
| 443 |
+
learning_rate=FLAGS.learning_rate,
|
| 444 |
+
num_train_steps=FLAGS.num_train_steps,
|
| 445 |
+
num_warmup_steps=FLAGS.num_warmup_steps,
|
| 446 |
+
use_tpu=FLAGS.use_tpu,
|
| 447 |
+
use_one_hot_embeddings=FLAGS.use_tpu)
|
| 448 |
+
|
| 449 |
+
# If TPU is not available, this will fall back to normal Estimator on CPU
|
| 450 |
+
# or GPU.
|
| 451 |
+
estimator = tf.contrib.tpu.TPUEstimator(
|
| 452 |
+
use_tpu=FLAGS.use_tpu,
|
| 453 |
+
model_fn=model_fn,
|
| 454 |
+
config=run_config,
|
| 455 |
+
train_batch_size=FLAGS.train_batch_size,
|
| 456 |
+
eval_batch_size=FLAGS.eval_batch_size)
|
| 457 |
+
|
| 458 |
+
if FLAGS.do_train:
|
| 459 |
+
tf.logging.info("***** Running training *****")
|
| 460 |
+
tf.logging.info(" Batch size = %d", FLAGS.train_batch_size)
|
| 461 |
+
train_input_fn = input_fn_builder(
|
| 462 |
+
input_files=input_files,
|
| 463 |
+
max_seq_length=FLAGS.max_seq_length,
|
| 464 |
+
max_predictions_per_seq=FLAGS.max_predictions_per_seq,
|
| 465 |
+
is_training=True)
|
| 466 |
+
estimator.train(input_fn=train_input_fn, max_steps=FLAGS.num_train_steps)
|
| 467 |
+
|
| 468 |
+
if FLAGS.do_eval:
|
| 469 |
+
tf.logging.info("***** Running evaluation *****")
|
| 470 |
+
tf.logging.info(" Batch size = %d", FLAGS.eval_batch_size)
|
| 471 |
+
|
| 472 |
+
eval_input_fn = input_fn_builder(
|
| 473 |
+
input_files=input_files,
|
| 474 |
+
max_seq_length=FLAGS.max_seq_length,
|
| 475 |
+
max_predictions_per_seq=FLAGS.max_predictions_per_seq,
|
| 476 |
+
is_training=False)
|
| 477 |
+
|
| 478 |
+
result = estimator.evaluate(
|
| 479 |
+
input_fn=eval_input_fn, steps=FLAGS.max_eval_steps)
|
| 480 |
+
|
| 481 |
+
output_eval_file = os.path.join(FLAGS.output_dir, "eval_results.txt")
|
| 482 |
+
with tf.gfile.GFile(output_eval_file, "w") as writer:
|
| 483 |
+
tf.logging.info("***** Eval results *****")
|
| 484 |
+
for key in sorted(result.keys()):
|
| 485 |
+
tf.logging.info(" %s = %s", key, str(result[key]))
|
| 486 |
+
writer.write("%s = %s\n" % (key, str(result[key])))
|
| 487 |
+
|
| 488 |
+
|
| 489 |
+
if __name__ == "__main__":
|
| 490 |
+
flags.mark_flag_as_required("input_file")
|
| 491 |
+
flags.mark_flag_as_required("bert_config_file")
|
| 492 |
+
flags.mark_flag_as_required("output_dir")
|
| 493 |
+
tf.app.run()
|
bert-master/bert-master/run_squad.py
ADDED
|
@@ -0,0 +1,1283 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# coding=utf-8
|
| 2 |
+
# Copyright 2018 The Google AI Language Team Authors.
|
| 3 |
+
#
|
| 4 |
+
# Licensed under the Apache License, Version 2.0 (the "License");
|
| 5 |
+
# you may not use this file except in compliance with the License.
|
| 6 |
+
# You may obtain a copy of the License at
|
| 7 |
+
#
|
| 8 |
+
# http://www.apache.org/licenses/LICENSE-2.0
|
| 9 |
+
#
|
| 10 |
+
# Unless required by applicable law or agreed to in writing, software
|
| 11 |
+
# distributed under the License is distributed on an "AS IS" BASIS,
|
| 12 |
+
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
| 13 |
+
# See the License for the specific language governing permissions and
|
| 14 |
+
# limitations under the License.
|
| 15 |
+
"""Run BERT on SQuAD 1.1 and SQuAD 2.0."""
|
| 16 |
+
|
| 17 |
+
from __future__ import absolute_import
|
| 18 |
+
from __future__ import division
|
| 19 |
+
from __future__ import print_function
|
| 20 |
+
|
| 21 |
+
import collections
|
| 22 |
+
import json
|
| 23 |
+
import math
|
| 24 |
+
import os
|
| 25 |
+
import random
|
| 26 |
+
import modeling
|
| 27 |
+
import optimization
|
| 28 |
+
import tokenization
|
| 29 |
+
import six
|
| 30 |
+
import tensorflow as tf
|
| 31 |
+
|
| 32 |
+
flags = tf.flags
|
| 33 |
+
|
| 34 |
+
FLAGS = flags.FLAGS
|
| 35 |
+
|
| 36 |
+
## Required parameters
|
| 37 |
+
flags.DEFINE_string(
|
| 38 |
+
"bert_config_file", None,
|
| 39 |
+
"The config json file corresponding to the pre-trained BERT model. "
|
| 40 |
+
"This specifies the model architecture.")
|
| 41 |
+
|
| 42 |
+
flags.DEFINE_string("vocab_file", None,
|
| 43 |
+
"The vocabulary file that the BERT model was trained on.")
|
| 44 |
+
|
| 45 |
+
flags.DEFINE_string(
|
| 46 |
+
"output_dir", None,
|
| 47 |
+
"The output directory where the model checkpoints will be written.")
|
| 48 |
+
|
| 49 |
+
## Other parameters
|
| 50 |
+
flags.DEFINE_string("train_file", None,
|
| 51 |
+
"SQuAD json for training. E.g., train-v1.1.json")
|
| 52 |
+
|
| 53 |
+
flags.DEFINE_string(
|
| 54 |
+
"predict_file", None,
|
| 55 |
+
"SQuAD json for predictions. E.g., dev-v1.1.json or test-v1.1.json")
|
| 56 |
+
|
| 57 |
+
flags.DEFINE_string(
|
| 58 |
+
"init_checkpoint", None,
|
| 59 |
+
"Initial checkpoint (usually from a pre-trained BERT model).")
|
| 60 |
+
|
| 61 |
+
flags.DEFINE_bool(
|
| 62 |
+
"do_lower_case", True,
|
| 63 |
+
"Whether to lower case the input text. Should be True for uncased "
|
| 64 |
+
"models and False for cased models.")
|
| 65 |
+
|
| 66 |
+
flags.DEFINE_integer(
|
| 67 |
+
"max_seq_length", 384,
|
| 68 |
+
"The maximum total input sequence length after WordPiece tokenization. "
|
| 69 |
+
"Sequences longer than this will be truncated, and sequences shorter "
|
| 70 |
+
"than this will be padded.")
|
| 71 |
+
|
| 72 |
+
flags.DEFINE_integer(
|
| 73 |
+
"doc_stride", 128,
|
| 74 |
+
"When splitting up a long document into chunks, how much stride to "
|
| 75 |
+
"take between chunks.")
|
| 76 |
+
|
| 77 |
+
flags.DEFINE_integer(
|
| 78 |
+
"max_query_length", 64,
|
| 79 |
+
"The maximum number of tokens for the question. Questions longer than "
|
| 80 |
+
"this will be truncated to this length.")
|
| 81 |
+
|
| 82 |
+
flags.DEFINE_bool("do_train", False, "Whether to run training.")
|
| 83 |
+
|
| 84 |
+
flags.DEFINE_bool("do_predict", False, "Whether to run eval on the dev set.")
|
| 85 |
+
|
| 86 |
+
flags.DEFINE_integer("train_batch_size", 32, "Total batch size for training.")
|
| 87 |
+
|
| 88 |
+
flags.DEFINE_integer("predict_batch_size", 8,
|
| 89 |
+
"Total batch size for predictions.")
|
| 90 |
+
|
| 91 |
+
flags.DEFINE_float("learning_rate", 5e-5, "The initial learning rate for Adam.")
|
| 92 |
+
|
| 93 |
+
flags.DEFINE_float("num_train_epochs", 3.0,
|
| 94 |
+
"Total number of training epochs to perform.")
|
| 95 |
+
|
| 96 |
+
flags.DEFINE_float(
|
| 97 |
+
"warmup_proportion", 0.1,
|
| 98 |
+
"Proportion of training to perform linear learning rate warmup for. "
|
| 99 |
+
"E.g., 0.1 = 10% of training.")
|
| 100 |
+
|
| 101 |
+
flags.DEFINE_integer("save_checkpoints_steps", 1000,
|
| 102 |
+
"How often to save the model checkpoint.")
|
| 103 |
+
|
| 104 |
+
flags.DEFINE_integer("iterations_per_loop", 1000,
|
| 105 |
+
"How many steps to make in each estimator call.")
|
| 106 |
+
|
| 107 |
+
flags.DEFINE_integer(
|
| 108 |
+
"n_best_size", 20,
|
| 109 |
+
"The total number of n-best predictions to generate in the "
|
| 110 |
+
"nbest_predictions.json output file.")
|
| 111 |
+
|
| 112 |
+
flags.DEFINE_integer(
|
| 113 |
+
"max_answer_length", 30,
|
| 114 |
+
"The maximum length of an answer that can be generated. This is needed "
|
| 115 |
+
"because the start and end predictions are not conditioned on one another.")
|
| 116 |
+
|
| 117 |
+
flags.DEFINE_bool("use_tpu", False, "Whether to use TPU or GPU/CPU.")
|
| 118 |
+
|
| 119 |
+
tf.flags.DEFINE_string(
|
| 120 |
+
"tpu_name", None,
|
| 121 |
+
"The Cloud TPU to use for training. This should be either the name "
|
| 122 |
+
"used when creating the Cloud TPU, or a grpc://ip.address.of.tpu:8470 "
|
| 123 |
+
"url.")
|
| 124 |
+
|
| 125 |
+
tf.flags.DEFINE_string(
|
| 126 |
+
"tpu_zone", None,
|
| 127 |
+
"[Optional] GCE zone where the Cloud TPU is located in. If not "
|
| 128 |
+
"specified, we will attempt to automatically detect the GCE project from "
|
| 129 |
+
"metadata.")
|
| 130 |
+
|
| 131 |
+
tf.flags.DEFINE_string(
|
| 132 |
+
"gcp_project", None,
|
| 133 |
+
"[Optional] Project name for the Cloud TPU-enabled project. If not "
|
| 134 |
+
"specified, we will attempt to automatically detect the GCE project from "
|
| 135 |
+
"metadata.")
|
| 136 |
+
|
| 137 |
+
tf.flags.DEFINE_string("master", None, "[Optional] TensorFlow master URL.")
|
| 138 |
+
|
| 139 |
+
flags.DEFINE_integer(
|
| 140 |
+
"num_tpu_cores", 8,
|
| 141 |
+
"Only used if `use_tpu` is True. Total number of TPU cores to use.")
|
| 142 |
+
|
| 143 |
+
flags.DEFINE_bool(
|
| 144 |
+
"verbose_logging", False,
|
| 145 |
+
"If true, all of the warnings related to data processing will be printed. "
|
| 146 |
+
"A number of warnings are expected for a normal SQuAD evaluation.")
|
| 147 |
+
|
| 148 |
+
flags.DEFINE_bool(
|
| 149 |
+
"version_2_with_negative", False,
|
| 150 |
+
"If true, the SQuAD examples contain some that do not have an answer.")
|
| 151 |
+
|
| 152 |
+
flags.DEFINE_float(
|
| 153 |
+
"null_score_diff_threshold", 0.0,
|
| 154 |
+
"If null_score - best_non_null is greater than the threshold predict null.")
|
| 155 |
+
|
| 156 |
+
|
| 157 |
+
class SquadExample(object):
|
| 158 |
+
"""A single training/test example for simple sequence classification.
|
| 159 |
+
|
| 160 |
+
For examples without an answer, the start and end position are -1.
|
| 161 |
+
"""
|
| 162 |
+
|
| 163 |
+
def __init__(self,
|
| 164 |
+
qas_id,
|
| 165 |
+
question_text,
|
| 166 |
+
doc_tokens,
|
| 167 |
+
orig_answer_text=None,
|
| 168 |
+
start_position=None,
|
| 169 |
+
end_position=None,
|
| 170 |
+
is_impossible=False):
|
| 171 |
+
self.qas_id = qas_id
|
| 172 |
+
self.question_text = question_text
|
| 173 |
+
self.doc_tokens = doc_tokens
|
| 174 |
+
self.orig_answer_text = orig_answer_text
|
| 175 |
+
self.start_position = start_position
|
| 176 |
+
self.end_position = end_position
|
| 177 |
+
self.is_impossible = is_impossible
|
| 178 |
+
|
| 179 |
+
def __str__(self):
|
| 180 |
+
return self.__repr__()
|
| 181 |
+
|
| 182 |
+
def __repr__(self):
|
| 183 |
+
s = ""
|
| 184 |
+
s += "qas_id: %s" % (tokenization.printable_text(self.qas_id))
|
| 185 |
+
s += ", question_text: %s" % (
|
| 186 |
+
tokenization.printable_text(self.question_text))
|
| 187 |
+
s += ", doc_tokens: [%s]" % (" ".join(self.doc_tokens))
|
| 188 |
+
if self.start_position:
|
| 189 |
+
s += ", start_position: %d" % (self.start_position)
|
| 190 |
+
if self.start_position:
|
| 191 |
+
s += ", end_position: %d" % (self.end_position)
|
| 192 |
+
if self.start_position:
|
| 193 |
+
s += ", is_impossible: %r" % (self.is_impossible)
|
| 194 |
+
return s
|
| 195 |
+
|
| 196 |
+
|
| 197 |
+
class InputFeatures(object):
|
| 198 |
+
"""A single set of features of data."""
|
| 199 |
+
|
| 200 |
+
def __init__(self,
|
| 201 |
+
unique_id,
|
| 202 |
+
example_index,
|
| 203 |
+
doc_span_index,
|
| 204 |
+
tokens,
|
| 205 |
+
token_to_orig_map,
|
| 206 |
+
token_is_max_context,
|
| 207 |
+
input_ids,
|
| 208 |
+
input_mask,
|
| 209 |
+
segment_ids,
|
| 210 |
+
start_position=None,
|
| 211 |
+
end_position=None,
|
| 212 |
+
is_impossible=None):
|
| 213 |
+
self.unique_id = unique_id
|
| 214 |
+
self.example_index = example_index
|
| 215 |
+
self.doc_span_index = doc_span_index
|
| 216 |
+
self.tokens = tokens
|
| 217 |
+
self.token_to_orig_map = token_to_orig_map
|
| 218 |
+
self.token_is_max_context = token_is_max_context
|
| 219 |
+
self.input_ids = input_ids
|
| 220 |
+
self.input_mask = input_mask
|
| 221 |
+
self.segment_ids = segment_ids
|
| 222 |
+
self.start_position = start_position
|
| 223 |
+
self.end_position = end_position
|
| 224 |
+
self.is_impossible = is_impossible
|
| 225 |
+
|
| 226 |
+
|
| 227 |
+
def read_squad_examples(input_file, is_training):
|
| 228 |
+
"""Read a SQuAD json file into a list of SquadExample."""
|
| 229 |
+
with tf.gfile.Open(input_file, "r") as reader:
|
| 230 |
+
input_data = json.load(reader)["data"]
|
| 231 |
+
|
| 232 |
+
def is_whitespace(c):
|
| 233 |
+
if c == " " or c == "\t" or c == "\r" or c == "\n" or ord(c) == 0x202F:
|
| 234 |
+
return True
|
| 235 |
+
return False
|
| 236 |
+
|
| 237 |
+
examples = []
|
| 238 |
+
for entry in input_data:
|
| 239 |
+
for paragraph in entry["paragraphs"]:
|
| 240 |
+
paragraph_text = paragraph["context"]
|
| 241 |
+
doc_tokens = []
|
| 242 |
+
char_to_word_offset = []
|
| 243 |
+
prev_is_whitespace = True
|
| 244 |
+
for c in paragraph_text:
|
| 245 |
+
if is_whitespace(c):
|
| 246 |
+
prev_is_whitespace = True
|
| 247 |
+
else:
|
| 248 |
+
if prev_is_whitespace:
|
| 249 |
+
doc_tokens.append(c)
|
| 250 |
+
else:
|
| 251 |
+
doc_tokens[-1] += c
|
| 252 |
+
prev_is_whitespace = False
|
| 253 |
+
char_to_word_offset.append(len(doc_tokens) - 1)
|
| 254 |
+
|
| 255 |
+
for qa in paragraph["qas"]:
|
| 256 |
+
qas_id = qa["id"]
|
| 257 |
+
question_text = qa["question"]
|
| 258 |
+
start_position = None
|
| 259 |
+
end_position = None
|
| 260 |
+
orig_answer_text = None
|
| 261 |
+
is_impossible = False
|
| 262 |
+
if is_training:
|
| 263 |
+
|
| 264 |
+
if FLAGS.version_2_with_negative:
|
| 265 |
+
is_impossible = qa["is_impossible"]
|
| 266 |
+
if (len(qa["answers"]) != 1) and (not is_impossible):
|
| 267 |
+
raise ValueError(
|
| 268 |
+
"For training, each question should have exactly 1 answer.")
|
| 269 |
+
if not is_impossible:
|
| 270 |
+
answer = qa["answers"][0]
|
| 271 |
+
orig_answer_text = answer["text"]
|
| 272 |
+
answer_offset = answer["answer_start"]
|
| 273 |
+
answer_length = len(orig_answer_text)
|
| 274 |
+
start_position = char_to_word_offset[answer_offset]
|
| 275 |
+
end_position = char_to_word_offset[answer_offset + answer_length -
|
| 276 |
+
1]
|
| 277 |
+
# Only add answers where the text can be exactly recovered from the
|
| 278 |
+
# document. If this CAN'T happen it's likely due to weird Unicode
|
| 279 |
+
# stuff so we will just skip the example.
|
| 280 |
+
#
|
| 281 |
+
# Note that this means for training mode, every example is NOT
|
| 282 |
+
# guaranteed to be preserved.
|
| 283 |
+
actual_text = " ".join(
|
| 284 |
+
doc_tokens[start_position:(end_position + 1)])
|
| 285 |
+
cleaned_answer_text = " ".join(
|
| 286 |
+
tokenization.whitespace_tokenize(orig_answer_text))
|
| 287 |
+
if actual_text.find(cleaned_answer_text) == -1:
|
| 288 |
+
tf.logging.warning("Could not find answer: '%s' vs. '%s'",
|
| 289 |
+
actual_text, cleaned_answer_text)
|
| 290 |
+
continue
|
| 291 |
+
else:
|
| 292 |
+
start_position = -1
|
| 293 |
+
end_position = -1
|
| 294 |
+
orig_answer_text = ""
|
| 295 |
+
|
| 296 |
+
example = SquadExample(
|
| 297 |
+
qas_id=qas_id,
|
| 298 |
+
question_text=question_text,
|
| 299 |
+
doc_tokens=doc_tokens,
|
| 300 |
+
orig_answer_text=orig_answer_text,
|
| 301 |
+
start_position=start_position,
|
| 302 |
+
end_position=end_position,
|
| 303 |
+
is_impossible=is_impossible)
|
| 304 |
+
examples.append(example)
|
| 305 |
+
|
| 306 |
+
return examples
|
| 307 |
+
|
| 308 |
+
|
| 309 |
+
def convert_examples_to_features(examples, tokenizer, max_seq_length,
|
| 310 |
+
doc_stride, max_query_length, is_training,
|
| 311 |
+
output_fn):
|
| 312 |
+
"""Loads a data file into a list of `InputBatch`s."""
|
| 313 |
+
|
| 314 |
+
unique_id = 1000000000
|
| 315 |
+
|
| 316 |
+
for (example_index, example) in enumerate(examples):
|
| 317 |
+
query_tokens = tokenizer.tokenize(example.question_text)
|
| 318 |
+
|
| 319 |
+
if len(query_tokens) > max_query_length:
|
| 320 |
+
query_tokens = query_tokens[0:max_query_length]
|
| 321 |
+
|
| 322 |
+
tok_to_orig_index = []
|
| 323 |
+
orig_to_tok_index = []
|
| 324 |
+
all_doc_tokens = []
|
| 325 |
+
for (i, token) in enumerate(example.doc_tokens):
|
| 326 |
+
orig_to_tok_index.append(len(all_doc_tokens))
|
| 327 |
+
sub_tokens = tokenizer.tokenize(token)
|
| 328 |
+
for sub_token in sub_tokens:
|
| 329 |
+
tok_to_orig_index.append(i)
|
| 330 |
+
all_doc_tokens.append(sub_token)
|
| 331 |
+
|
| 332 |
+
tok_start_position = None
|
| 333 |
+
tok_end_position = None
|
| 334 |
+
if is_training and example.is_impossible:
|
| 335 |
+
tok_start_position = -1
|
| 336 |
+
tok_end_position = -1
|
| 337 |
+
if is_training and not example.is_impossible:
|
| 338 |
+
tok_start_position = orig_to_tok_index[example.start_position]
|
| 339 |
+
if example.end_position < len(example.doc_tokens) - 1:
|
| 340 |
+
tok_end_position = orig_to_tok_index[example.end_position + 1] - 1
|
| 341 |
+
else:
|
| 342 |
+
tok_end_position = len(all_doc_tokens) - 1
|
| 343 |
+
(tok_start_position, tok_end_position) = _improve_answer_span(
|
| 344 |
+
all_doc_tokens, tok_start_position, tok_end_position, tokenizer,
|
| 345 |
+
example.orig_answer_text)
|
| 346 |
+
|
| 347 |
+
# The -3 accounts for [CLS], [SEP] and [SEP]
|
| 348 |
+
max_tokens_for_doc = max_seq_length - len(query_tokens) - 3
|
| 349 |
+
|
| 350 |
+
# We can have documents that are longer than the maximum sequence length.
|
| 351 |
+
# To deal with this we do a sliding window approach, where we take chunks
|
| 352 |
+
# of the up to our max length with a stride of `doc_stride`.
|
| 353 |
+
_DocSpan = collections.namedtuple( # pylint: disable=invalid-name
|
| 354 |
+
"DocSpan", ["start", "length"])
|
| 355 |
+
doc_spans = []
|
| 356 |
+
start_offset = 0
|
| 357 |
+
while start_offset < len(all_doc_tokens):
|
| 358 |
+
length = len(all_doc_tokens) - start_offset
|
| 359 |
+
if length > max_tokens_for_doc:
|
| 360 |
+
length = max_tokens_for_doc
|
| 361 |
+
doc_spans.append(_DocSpan(start=start_offset, length=length))
|
| 362 |
+
if start_offset + length == len(all_doc_tokens):
|
| 363 |
+
break
|
| 364 |
+
start_offset += min(length, doc_stride)
|
| 365 |
+
|
| 366 |
+
for (doc_span_index, doc_span) in enumerate(doc_spans):
|
| 367 |
+
tokens = []
|
| 368 |
+
token_to_orig_map = {}
|
| 369 |
+
token_is_max_context = {}
|
| 370 |
+
segment_ids = []
|
| 371 |
+
tokens.append("[CLS]")
|
| 372 |
+
segment_ids.append(0)
|
| 373 |
+
for token in query_tokens:
|
| 374 |
+
tokens.append(token)
|
| 375 |
+
segment_ids.append(0)
|
| 376 |
+
tokens.append("[SEP]")
|
| 377 |
+
segment_ids.append(0)
|
| 378 |
+
|
| 379 |
+
for i in range(doc_span.length):
|
| 380 |
+
split_token_index = doc_span.start + i
|
| 381 |
+
token_to_orig_map[len(tokens)] = tok_to_orig_index[split_token_index]
|
| 382 |
+
|
| 383 |
+
is_max_context = _check_is_max_context(doc_spans, doc_span_index,
|
| 384 |
+
split_token_index)
|
| 385 |
+
token_is_max_context[len(tokens)] = is_max_context
|
| 386 |
+
tokens.append(all_doc_tokens[split_token_index])
|
| 387 |
+
segment_ids.append(1)
|
| 388 |
+
tokens.append("[SEP]")
|
| 389 |
+
segment_ids.append(1)
|
| 390 |
+
|
| 391 |
+
input_ids = tokenizer.convert_tokens_to_ids(tokens)
|
| 392 |
+
|
| 393 |
+
# The mask has 1 for real tokens and 0 for padding tokens. Only real
|
| 394 |
+
# tokens are attended to.
|
| 395 |
+
input_mask = [1] * len(input_ids)
|
| 396 |
+
|
| 397 |
+
# Zero-pad up to the sequence length.
|
| 398 |
+
while len(input_ids) < max_seq_length:
|
| 399 |
+
input_ids.append(0)
|
| 400 |
+
input_mask.append(0)
|
| 401 |
+
segment_ids.append(0)
|
| 402 |
+
|
| 403 |
+
assert len(input_ids) == max_seq_length
|
| 404 |
+
assert len(input_mask) == max_seq_length
|
| 405 |
+
assert len(segment_ids) == max_seq_length
|
| 406 |
+
|
| 407 |
+
start_position = None
|
| 408 |
+
end_position = None
|
| 409 |
+
if is_training and not example.is_impossible:
|
| 410 |
+
# For training, if our document chunk does not contain an annotation
|
| 411 |
+
# we throw it out, since there is nothing to predict.
|
| 412 |
+
doc_start = doc_span.start
|
| 413 |
+
doc_end = doc_span.start + doc_span.length - 1
|
| 414 |
+
out_of_span = False
|
| 415 |
+
if not (tok_start_position >= doc_start and
|
| 416 |
+
tok_end_position <= doc_end):
|
| 417 |
+
out_of_span = True
|
| 418 |
+
if out_of_span:
|
| 419 |
+
start_position = 0
|
| 420 |
+
end_position = 0
|
| 421 |
+
else:
|
| 422 |
+
doc_offset = len(query_tokens) + 2
|
| 423 |
+
start_position = tok_start_position - doc_start + doc_offset
|
| 424 |
+
end_position = tok_end_position - doc_start + doc_offset
|
| 425 |
+
|
| 426 |
+
if is_training and example.is_impossible:
|
| 427 |
+
start_position = 0
|
| 428 |
+
end_position = 0
|
| 429 |
+
|
| 430 |
+
if example_index < 20:
|
| 431 |
+
tf.logging.info("*** Example ***")
|
| 432 |
+
tf.logging.info("unique_id: %s" % (unique_id))
|
| 433 |
+
tf.logging.info("example_index: %s" % (example_index))
|
| 434 |
+
tf.logging.info("doc_span_index: %s" % (doc_span_index))
|
| 435 |
+
tf.logging.info("tokens: %s" % " ".join(
|
| 436 |
+
[tokenization.printable_text(x) for x in tokens]))
|
| 437 |
+
tf.logging.info("token_to_orig_map: %s" % " ".join(
|
| 438 |
+
["%d:%d" % (x, y) for (x, y) in six.iteritems(token_to_orig_map)]))
|
| 439 |
+
tf.logging.info("token_is_max_context: %s" % " ".join([
|
| 440 |
+
"%d:%s" % (x, y) for (x, y) in six.iteritems(token_is_max_context)
|
| 441 |
+
]))
|
| 442 |
+
tf.logging.info("input_ids: %s" % " ".join([str(x) for x in input_ids]))
|
| 443 |
+
tf.logging.info(
|
| 444 |
+
"input_mask: %s" % " ".join([str(x) for x in input_mask]))
|
| 445 |
+
tf.logging.info(
|
| 446 |
+
"segment_ids: %s" % " ".join([str(x) for x in segment_ids]))
|
| 447 |
+
if is_training and example.is_impossible:
|
| 448 |
+
tf.logging.info("impossible example")
|
| 449 |
+
if is_training and not example.is_impossible:
|
| 450 |
+
answer_text = " ".join(tokens[start_position:(end_position + 1)])
|
| 451 |
+
tf.logging.info("start_position: %d" % (start_position))
|
| 452 |
+
tf.logging.info("end_position: %d" % (end_position))
|
| 453 |
+
tf.logging.info(
|
| 454 |
+
"answer: %s" % (tokenization.printable_text(answer_text)))
|
| 455 |
+
|
| 456 |
+
feature = InputFeatures(
|
| 457 |
+
unique_id=unique_id,
|
| 458 |
+
example_index=example_index,
|
| 459 |
+
doc_span_index=doc_span_index,
|
| 460 |
+
tokens=tokens,
|
| 461 |
+
token_to_orig_map=token_to_orig_map,
|
| 462 |
+
token_is_max_context=token_is_max_context,
|
| 463 |
+
input_ids=input_ids,
|
| 464 |
+
input_mask=input_mask,
|
| 465 |
+
segment_ids=segment_ids,
|
| 466 |
+
start_position=start_position,
|
| 467 |
+
end_position=end_position,
|
| 468 |
+
is_impossible=example.is_impossible)
|
| 469 |
+
|
| 470 |
+
# Run callback
|
| 471 |
+
output_fn(feature)
|
| 472 |
+
|
| 473 |
+
unique_id += 1
|
| 474 |
+
|
| 475 |
+
|
| 476 |
+
def _improve_answer_span(doc_tokens, input_start, input_end, tokenizer,
|
| 477 |
+
orig_answer_text):
|
| 478 |
+
"""Returns tokenized answer spans that better match the annotated answer."""
|
| 479 |
+
|
| 480 |
+
# The SQuAD annotations are character based. We first project them to
|
| 481 |
+
# whitespace-tokenized words. But then after WordPiece tokenization, we can
|
| 482 |
+
# often find a "better match". For example:
|
| 483 |
+
#
|
| 484 |
+
# Question: What year was John Smith born?
|
| 485 |
+
# Context: The leader was John Smith (1895-1943).
|
| 486 |
+
# Answer: 1895
|
| 487 |
+
#
|
| 488 |
+
# The original whitespace-tokenized answer will be "(1895-1943).". However
|
| 489 |
+
# after tokenization, our tokens will be "( 1895 - 1943 ) .". So we can match
|
| 490 |
+
# the exact answer, 1895.
|
| 491 |
+
#
|
| 492 |
+
# However, this is not always possible. Consider the following:
|
| 493 |
+
#
|
| 494 |
+
# Question: What country is the top exporter of electornics?
|
| 495 |
+
# Context: The Japanese electronics industry is the lagest in the world.
|
| 496 |
+
# Answer: Japan
|
| 497 |
+
#
|
| 498 |
+
# In this case, the annotator chose "Japan" as a character sub-span of
|
| 499 |
+
# the word "Japanese". Since our WordPiece tokenizer does not split
|
| 500 |
+
# "Japanese", we just use "Japanese" as the annotation. This is fairly rare
|
| 501 |
+
# in SQuAD, but does happen.
|
| 502 |
+
tok_answer_text = " ".join(tokenizer.tokenize(orig_answer_text))
|
| 503 |
+
|
| 504 |
+
for new_start in range(input_start, input_end + 1):
|
| 505 |
+
for new_end in range(input_end, new_start - 1, -1):
|
| 506 |
+
text_span = " ".join(doc_tokens[new_start:(new_end + 1)])
|
| 507 |
+
if text_span == tok_answer_text:
|
| 508 |
+
return (new_start, new_end)
|
| 509 |
+
|
| 510 |
+
return (input_start, input_end)
|
| 511 |
+
|
| 512 |
+
|
| 513 |
+
def _check_is_max_context(doc_spans, cur_span_index, position):
|
| 514 |
+
"""Check if this is the 'max context' doc span for the token."""
|
| 515 |
+
|
| 516 |
+
# Because of the sliding window approach taken to scoring documents, a single
|
| 517 |
+
# token can appear in multiple documents. E.g.
|
| 518 |
+
# Doc: the man went to the store and bought a gallon of milk
|
| 519 |
+
# Span A: the man went to the
|
| 520 |
+
# Span B: to the store and bought
|
| 521 |
+
# Span C: and bought a gallon of
|
| 522 |
+
# ...
|
| 523 |
+
#
|
| 524 |
+
# Now the word 'bought' will have two scores from spans B and C. We only
|
| 525 |
+
# want to consider the score with "maximum context", which we define as
|
| 526 |
+
# the *minimum* of its left and right context (the *sum* of left and
|
| 527 |
+
# right context will always be the same, of course).
|
| 528 |
+
#
|
| 529 |
+
# In the example the maximum context for 'bought' would be span C since
|
| 530 |
+
# it has 1 left context and 3 right context, while span B has 4 left context
|
| 531 |
+
# and 0 right context.
|
| 532 |
+
best_score = None
|
| 533 |
+
best_span_index = None
|
| 534 |
+
for (span_index, doc_span) in enumerate(doc_spans):
|
| 535 |
+
end = doc_span.start + doc_span.length - 1
|
| 536 |
+
if position < doc_span.start:
|
| 537 |
+
continue
|
| 538 |
+
if position > end:
|
| 539 |
+
continue
|
| 540 |
+
num_left_context = position - doc_span.start
|
| 541 |
+
num_right_context = end - position
|
| 542 |
+
score = min(num_left_context, num_right_context) + 0.01 * doc_span.length
|
| 543 |
+
if best_score is None or score > best_score:
|
| 544 |
+
best_score = score
|
| 545 |
+
best_span_index = span_index
|
| 546 |
+
|
| 547 |
+
return cur_span_index == best_span_index
|
| 548 |
+
|
| 549 |
+
|
| 550 |
+
def create_model(bert_config, is_training, input_ids, input_mask, segment_ids,
|
| 551 |
+
use_one_hot_embeddings):
|
| 552 |
+
"""Creates a classification model."""
|
| 553 |
+
model = modeling.BertModel(
|
| 554 |
+
config=bert_config,
|
| 555 |
+
is_training=is_training,
|
| 556 |
+
input_ids=input_ids,
|
| 557 |
+
input_mask=input_mask,
|
| 558 |
+
token_type_ids=segment_ids,
|
| 559 |
+
use_one_hot_embeddings=use_one_hot_embeddings)
|
| 560 |
+
|
| 561 |
+
final_hidden = model.get_sequence_output()
|
| 562 |
+
|
| 563 |
+
final_hidden_shape = modeling.get_shape_list(final_hidden, expected_rank=3)
|
| 564 |
+
batch_size = final_hidden_shape[0]
|
| 565 |
+
seq_length = final_hidden_shape[1]
|
| 566 |
+
hidden_size = final_hidden_shape[2]
|
| 567 |
+
|
| 568 |
+
output_weights = tf.get_variable(
|
| 569 |
+
"cls/squad/output_weights", [2, hidden_size],
|
| 570 |
+
initializer=tf.truncated_normal_initializer(stddev=0.02))
|
| 571 |
+
|
| 572 |
+
output_bias = tf.get_variable(
|
| 573 |
+
"cls/squad/output_bias", [2], initializer=tf.zeros_initializer())
|
| 574 |
+
|
| 575 |
+
final_hidden_matrix = tf.reshape(final_hidden,
|
| 576 |
+
[batch_size * seq_length, hidden_size])
|
| 577 |
+
logits = tf.matmul(final_hidden_matrix, output_weights, transpose_b=True)
|
| 578 |
+
logits = tf.nn.bias_add(logits, output_bias)
|
| 579 |
+
|
| 580 |
+
logits = tf.reshape(logits, [batch_size, seq_length, 2])
|
| 581 |
+
logits = tf.transpose(logits, [2, 0, 1])
|
| 582 |
+
|
| 583 |
+
unstacked_logits = tf.unstack(logits, axis=0)
|
| 584 |
+
|
| 585 |
+
(start_logits, end_logits) = (unstacked_logits[0], unstacked_logits[1])
|
| 586 |
+
|
| 587 |
+
return (start_logits, end_logits)
|
| 588 |
+
|
| 589 |
+
|
| 590 |
+
def model_fn_builder(bert_config, init_checkpoint, learning_rate,
|
| 591 |
+
num_train_steps, num_warmup_steps, use_tpu,
|
| 592 |
+
use_one_hot_embeddings):
|
| 593 |
+
"""Returns `model_fn` closure for TPUEstimator."""
|
| 594 |
+
|
| 595 |
+
def model_fn(features, labels, mode, params): # pylint: disable=unused-argument
|
| 596 |
+
"""The `model_fn` for TPUEstimator."""
|
| 597 |
+
|
| 598 |
+
tf.logging.info("*** Features ***")
|
| 599 |
+
for name in sorted(features.keys()):
|
| 600 |
+
tf.logging.info(" name = %s, shape = %s" % (name, features[name].shape))
|
| 601 |
+
|
| 602 |
+
unique_ids = features["unique_ids"]
|
| 603 |
+
input_ids = features["input_ids"]
|
| 604 |
+
input_mask = features["input_mask"]
|
| 605 |
+
segment_ids = features["segment_ids"]
|
| 606 |
+
|
| 607 |
+
is_training = (mode == tf.estimator.ModeKeys.TRAIN)
|
| 608 |
+
|
| 609 |
+
(start_logits, end_logits) = create_model(
|
| 610 |
+
bert_config=bert_config,
|
| 611 |
+
is_training=is_training,
|
| 612 |
+
input_ids=input_ids,
|
| 613 |
+
input_mask=input_mask,
|
| 614 |
+
segment_ids=segment_ids,
|
| 615 |
+
use_one_hot_embeddings=use_one_hot_embeddings)
|
| 616 |
+
|
| 617 |
+
tvars = tf.trainable_variables()
|
| 618 |
+
|
| 619 |
+
initialized_variable_names = {}
|
| 620 |
+
scaffold_fn = None
|
| 621 |
+
if init_checkpoint:
|
| 622 |
+
(assignment_map, initialized_variable_names
|
| 623 |
+
) = modeling.get_assignment_map_from_checkpoint(tvars, init_checkpoint)
|
| 624 |
+
if use_tpu:
|
| 625 |
+
|
| 626 |
+
def tpu_scaffold():
|
| 627 |
+
tf.train.init_from_checkpoint(init_checkpoint, assignment_map)
|
| 628 |
+
return tf.train.Scaffold()
|
| 629 |
+
|
| 630 |
+
scaffold_fn = tpu_scaffold
|
| 631 |
+
else:
|
| 632 |
+
tf.train.init_from_checkpoint(init_checkpoint, assignment_map)
|
| 633 |
+
|
| 634 |
+
tf.logging.info("**** Trainable Variables ****")
|
| 635 |
+
for var in tvars:
|
| 636 |
+
init_string = ""
|
| 637 |
+
if var.name in initialized_variable_names:
|
| 638 |
+
init_string = ", *INIT_FROM_CKPT*"
|
| 639 |
+
tf.logging.info(" name = %s, shape = %s%s", var.name, var.shape,
|
| 640 |
+
init_string)
|
| 641 |
+
|
| 642 |
+
output_spec = None
|
| 643 |
+
if mode == tf.estimator.ModeKeys.TRAIN:
|
| 644 |
+
seq_length = modeling.get_shape_list(input_ids)[1]
|
| 645 |
+
|
| 646 |
+
def compute_loss(logits, positions):
|
| 647 |
+
one_hot_positions = tf.one_hot(
|
| 648 |
+
positions, depth=seq_length, dtype=tf.float32)
|
| 649 |
+
log_probs = tf.nn.log_softmax(logits, axis=-1)
|
| 650 |
+
loss = -tf.reduce_mean(
|
| 651 |
+
tf.reduce_sum(one_hot_positions * log_probs, axis=-1))
|
| 652 |
+
return loss
|
| 653 |
+
|
| 654 |
+
start_positions = features["start_positions"]
|
| 655 |
+
end_positions = features["end_positions"]
|
| 656 |
+
|
| 657 |
+
start_loss = compute_loss(start_logits, start_positions)
|
| 658 |
+
end_loss = compute_loss(end_logits, end_positions)
|
| 659 |
+
|
| 660 |
+
total_loss = (start_loss + end_loss) / 2.0
|
| 661 |
+
|
| 662 |
+
train_op = optimization.create_optimizer(
|
| 663 |
+
total_loss, learning_rate, num_train_steps, num_warmup_steps, use_tpu)
|
| 664 |
+
|
| 665 |
+
output_spec = tf.contrib.tpu.TPUEstimatorSpec(
|
| 666 |
+
mode=mode,
|
| 667 |
+
loss=total_loss,
|
| 668 |
+
train_op=train_op,
|
| 669 |
+
scaffold_fn=scaffold_fn)
|
| 670 |
+
elif mode == tf.estimator.ModeKeys.PREDICT:
|
| 671 |
+
predictions = {
|
| 672 |
+
"unique_ids": unique_ids,
|
| 673 |
+
"start_logits": start_logits,
|
| 674 |
+
"end_logits": end_logits,
|
| 675 |
+
}
|
| 676 |
+
output_spec = tf.contrib.tpu.TPUEstimatorSpec(
|
| 677 |
+
mode=mode, predictions=predictions, scaffold_fn=scaffold_fn)
|
| 678 |
+
else:
|
| 679 |
+
raise ValueError(
|
| 680 |
+
"Only TRAIN and PREDICT modes are supported: %s" % (mode))
|
| 681 |
+
|
| 682 |
+
return output_spec
|
| 683 |
+
|
| 684 |
+
return model_fn
|
| 685 |
+
|
| 686 |
+
|
| 687 |
+
def input_fn_builder(input_file, seq_length, is_training, drop_remainder):
|
| 688 |
+
"""Creates an `input_fn` closure to be passed to TPUEstimator."""
|
| 689 |
+
|
| 690 |
+
name_to_features = {
|
| 691 |
+
"unique_ids": tf.FixedLenFeature([], tf.int64),
|
| 692 |
+
"input_ids": tf.FixedLenFeature([seq_length], tf.int64),
|
| 693 |
+
"input_mask": tf.FixedLenFeature([seq_length], tf.int64),
|
| 694 |
+
"segment_ids": tf.FixedLenFeature([seq_length], tf.int64),
|
| 695 |
+
}
|
| 696 |
+
|
| 697 |
+
if is_training:
|
| 698 |
+
name_to_features["start_positions"] = tf.FixedLenFeature([], tf.int64)
|
| 699 |
+
name_to_features["end_positions"] = tf.FixedLenFeature([], tf.int64)
|
| 700 |
+
|
| 701 |
+
def _decode_record(record, name_to_features):
|
| 702 |
+
"""Decodes a record to a TensorFlow example."""
|
| 703 |
+
example = tf.parse_single_example(record, name_to_features)
|
| 704 |
+
|
| 705 |
+
# tf.Example only supports tf.int64, but the TPU only supports tf.int32.
|
| 706 |
+
# So cast all int64 to int32.
|
| 707 |
+
for name in list(example.keys()):
|
| 708 |
+
t = example[name]
|
| 709 |
+
if t.dtype == tf.int64:
|
| 710 |
+
t = tf.to_int32(t)
|
| 711 |
+
example[name] = t
|
| 712 |
+
|
| 713 |
+
return example
|
| 714 |
+
|
| 715 |
+
def input_fn(params):
|
| 716 |
+
"""The actual input function."""
|
| 717 |
+
batch_size = params["batch_size"]
|
| 718 |
+
|
| 719 |
+
# For training, we want a lot of parallel reading and shuffling.
|
| 720 |
+
# For eval, we want no shuffling and parallel reading doesn't matter.
|
| 721 |
+
d = tf.data.TFRecordDataset(input_file)
|
| 722 |
+
if is_training:
|
| 723 |
+
d = d.repeat()
|
| 724 |
+
d = d.shuffle(buffer_size=100)
|
| 725 |
+
|
| 726 |
+
d = d.apply(
|
| 727 |
+
tf.contrib.data.map_and_batch(
|
| 728 |
+
lambda record: _decode_record(record, name_to_features),
|
| 729 |
+
batch_size=batch_size,
|
| 730 |
+
drop_remainder=drop_remainder))
|
| 731 |
+
|
| 732 |
+
return d
|
| 733 |
+
|
| 734 |
+
return input_fn
|
| 735 |
+
|
| 736 |
+
|
| 737 |
+
RawResult = collections.namedtuple("RawResult",
|
| 738 |
+
["unique_id", "start_logits", "end_logits"])
|
| 739 |
+
|
| 740 |
+
|
| 741 |
+
def write_predictions(all_examples, all_features, all_results, n_best_size,
|
| 742 |
+
max_answer_length, do_lower_case, output_prediction_file,
|
| 743 |
+
output_nbest_file, output_null_log_odds_file):
|
| 744 |
+
"""Write final predictions to the json file and log-odds of null if needed."""
|
| 745 |
+
tf.logging.info("Writing predictions to: %s" % (output_prediction_file))
|
| 746 |
+
tf.logging.info("Writing nbest to: %s" % (output_nbest_file))
|
| 747 |
+
|
| 748 |
+
example_index_to_features = collections.defaultdict(list)
|
| 749 |
+
for feature in all_features:
|
| 750 |
+
example_index_to_features[feature.example_index].append(feature)
|
| 751 |
+
|
| 752 |
+
unique_id_to_result = {}
|
| 753 |
+
for result in all_results:
|
| 754 |
+
unique_id_to_result[result.unique_id] = result
|
| 755 |
+
|
| 756 |
+
_PrelimPrediction = collections.namedtuple( # pylint: disable=invalid-name
|
| 757 |
+
"PrelimPrediction",
|
| 758 |
+
["feature_index", "start_index", "end_index", "start_logit", "end_logit"])
|
| 759 |
+
|
| 760 |
+
all_predictions = collections.OrderedDict()
|
| 761 |
+
all_nbest_json = collections.OrderedDict()
|
| 762 |
+
scores_diff_json = collections.OrderedDict()
|
| 763 |
+
|
| 764 |
+
for (example_index, example) in enumerate(all_examples):
|
| 765 |
+
features = example_index_to_features[example_index]
|
| 766 |
+
|
| 767 |
+
prelim_predictions = []
|
| 768 |
+
# keep track of the minimum score of null start+end of position 0
|
| 769 |
+
score_null = 1000000 # large and positive
|
| 770 |
+
min_null_feature_index = 0 # the paragraph slice with min mull score
|
| 771 |
+
null_start_logit = 0 # the start logit at the slice with min null score
|
| 772 |
+
null_end_logit = 0 # the end logit at the slice with min null score
|
| 773 |
+
for (feature_index, feature) in enumerate(features):
|
| 774 |
+
result = unique_id_to_result[feature.unique_id]
|
| 775 |
+
start_indexes = _get_best_indexes(result.start_logits, n_best_size)
|
| 776 |
+
end_indexes = _get_best_indexes(result.end_logits, n_best_size)
|
| 777 |
+
# if we could have irrelevant answers, get the min score of irrelevant
|
| 778 |
+
if FLAGS.version_2_with_negative:
|
| 779 |
+
feature_null_score = result.start_logits[0] + result.end_logits[0]
|
| 780 |
+
if feature_null_score < score_null:
|
| 781 |
+
score_null = feature_null_score
|
| 782 |
+
min_null_feature_index = feature_index
|
| 783 |
+
null_start_logit = result.start_logits[0]
|
| 784 |
+
null_end_logit = result.end_logits[0]
|
| 785 |
+
for start_index in start_indexes:
|
| 786 |
+
for end_index in end_indexes:
|
| 787 |
+
# We could hypothetically create invalid predictions, e.g., predict
|
| 788 |
+
# that the start of the span is in the question. We throw out all
|
| 789 |
+
# invalid predictions.
|
| 790 |
+
if start_index >= len(feature.tokens):
|
| 791 |
+
continue
|
| 792 |
+
if end_index >= len(feature.tokens):
|
| 793 |
+
continue
|
| 794 |
+
if start_index not in feature.token_to_orig_map:
|
| 795 |
+
continue
|
| 796 |
+
if end_index not in feature.token_to_orig_map:
|
| 797 |
+
continue
|
| 798 |
+
if not feature.token_is_max_context.get(start_index, False):
|
| 799 |
+
continue
|
| 800 |
+
if end_index < start_index:
|
| 801 |
+
continue
|
| 802 |
+
length = end_index - start_index + 1
|
| 803 |
+
if length > max_answer_length:
|
| 804 |
+
continue
|
| 805 |
+
prelim_predictions.append(
|
| 806 |
+
_PrelimPrediction(
|
| 807 |
+
feature_index=feature_index,
|
| 808 |
+
start_index=start_index,
|
| 809 |
+
end_index=end_index,
|
| 810 |
+
start_logit=result.start_logits[start_index],
|
| 811 |
+
end_logit=result.end_logits[end_index]))
|
| 812 |
+
|
| 813 |
+
if FLAGS.version_2_with_negative:
|
| 814 |
+
prelim_predictions.append(
|
| 815 |
+
_PrelimPrediction(
|
| 816 |
+
feature_index=min_null_feature_index,
|
| 817 |
+
start_index=0,
|
| 818 |
+
end_index=0,
|
| 819 |
+
start_logit=null_start_logit,
|
| 820 |
+
end_logit=null_end_logit))
|
| 821 |
+
prelim_predictions = sorted(
|
| 822 |
+
prelim_predictions,
|
| 823 |
+
key=lambda x: (x.start_logit + x.end_logit),
|
| 824 |
+
reverse=True)
|
| 825 |
+
|
| 826 |
+
_NbestPrediction = collections.namedtuple( # pylint: disable=invalid-name
|
| 827 |
+
"NbestPrediction", ["text", "start_logit", "end_logit"])
|
| 828 |
+
|
| 829 |
+
seen_predictions = {}
|
| 830 |
+
nbest = []
|
| 831 |
+
for pred in prelim_predictions:
|
| 832 |
+
if len(nbest) >= n_best_size:
|
| 833 |
+
break
|
| 834 |
+
feature = features[pred.feature_index]
|
| 835 |
+
if pred.start_index > 0: # this is a non-null prediction
|
| 836 |
+
tok_tokens = feature.tokens[pred.start_index:(pred.end_index + 1)]
|
| 837 |
+
orig_doc_start = feature.token_to_orig_map[pred.start_index]
|
| 838 |
+
orig_doc_end = feature.token_to_orig_map[pred.end_index]
|
| 839 |
+
orig_tokens = example.doc_tokens[orig_doc_start:(orig_doc_end + 1)]
|
| 840 |
+
tok_text = " ".join(tok_tokens)
|
| 841 |
+
|
| 842 |
+
# De-tokenize WordPieces that have been split off.
|
| 843 |
+
tok_text = tok_text.replace(" ##", "")
|
| 844 |
+
tok_text = tok_text.replace("##", "")
|
| 845 |
+
|
| 846 |
+
# Clean whitespace
|
| 847 |
+
tok_text = tok_text.strip()
|
| 848 |
+
tok_text = " ".join(tok_text.split())
|
| 849 |
+
orig_text = " ".join(orig_tokens)
|
| 850 |
+
|
| 851 |
+
final_text = get_final_text(tok_text, orig_text, do_lower_case)
|
| 852 |
+
if final_text in seen_predictions:
|
| 853 |
+
continue
|
| 854 |
+
|
| 855 |
+
seen_predictions[final_text] = True
|
| 856 |
+
else:
|
| 857 |
+
final_text = ""
|
| 858 |
+
seen_predictions[final_text] = True
|
| 859 |
+
|
| 860 |
+
nbest.append(
|
| 861 |
+
_NbestPrediction(
|
| 862 |
+
text=final_text,
|
| 863 |
+
start_logit=pred.start_logit,
|
| 864 |
+
end_logit=pred.end_logit))
|
| 865 |
+
|
| 866 |
+
# if we didn't inlude the empty option in the n-best, inlcude it
|
| 867 |
+
if FLAGS.version_2_with_negative:
|
| 868 |
+
if "" not in seen_predictions:
|
| 869 |
+
nbest.append(
|
| 870 |
+
_NbestPrediction(
|
| 871 |
+
text="", start_logit=null_start_logit,
|
| 872 |
+
end_logit=null_end_logit))
|
| 873 |
+
# In very rare edge cases we could have no valid predictions. So we
|
| 874 |
+
# just create a nonce prediction in this case to avoid failure.
|
| 875 |
+
if not nbest:
|
| 876 |
+
nbest.append(
|
| 877 |
+
_NbestPrediction(text="empty", start_logit=0.0, end_logit=0.0))
|
| 878 |
+
|
| 879 |
+
assert len(nbest) >= 1
|
| 880 |
+
|
| 881 |
+
total_scores = []
|
| 882 |
+
best_non_null_entry = None
|
| 883 |
+
for entry in nbest:
|
| 884 |
+
total_scores.append(entry.start_logit + entry.end_logit)
|
| 885 |
+
if not best_non_null_entry:
|
| 886 |
+
if entry.text:
|
| 887 |
+
best_non_null_entry = entry
|
| 888 |
+
|
| 889 |
+
probs = _compute_softmax(total_scores)
|
| 890 |
+
|
| 891 |
+
nbest_json = []
|
| 892 |
+
for (i, entry) in enumerate(nbest):
|
| 893 |
+
output = collections.OrderedDict()
|
| 894 |
+
output["text"] = entry.text
|
| 895 |
+
output["probability"] = probs[i]
|
| 896 |
+
output["start_logit"] = entry.start_logit
|
| 897 |
+
output["end_logit"] = entry.end_logit
|
| 898 |
+
nbest_json.append(output)
|
| 899 |
+
|
| 900 |
+
assert len(nbest_json) >= 1
|
| 901 |
+
|
| 902 |
+
if not FLAGS.version_2_with_negative:
|
| 903 |
+
all_predictions[example.qas_id] = nbest_json[0]["text"]
|
| 904 |
+
else:
|
| 905 |
+
# predict "" iff the null score - the score of best non-null > threshold
|
| 906 |
+
score_diff = score_null - best_non_null_entry.start_logit - (
|
| 907 |
+
best_non_null_entry.end_logit)
|
| 908 |
+
scores_diff_json[example.qas_id] = score_diff
|
| 909 |
+
if score_diff > FLAGS.null_score_diff_threshold:
|
| 910 |
+
all_predictions[example.qas_id] = ""
|
| 911 |
+
else:
|
| 912 |
+
all_predictions[example.qas_id] = best_non_null_entry.text
|
| 913 |
+
|
| 914 |
+
all_nbest_json[example.qas_id] = nbest_json
|
| 915 |
+
|
| 916 |
+
with tf.gfile.GFile(output_prediction_file, "w") as writer:
|
| 917 |
+
writer.write(json.dumps(all_predictions, indent=4) + "\n")
|
| 918 |
+
|
| 919 |
+
with tf.gfile.GFile(output_nbest_file, "w") as writer:
|
| 920 |
+
writer.write(json.dumps(all_nbest_json, indent=4) + "\n")
|
| 921 |
+
|
| 922 |
+
if FLAGS.version_2_with_negative:
|
| 923 |
+
with tf.gfile.GFile(output_null_log_odds_file, "w") as writer:
|
| 924 |
+
writer.write(json.dumps(scores_diff_json, indent=4) + "\n")
|
| 925 |
+
|
| 926 |
+
|
| 927 |
+
def get_final_text(pred_text, orig_text, do_lower_case):
|
| 928 |
+
"""Project the tokenized prediction back to the original text."""
|
| 929 |
+
|
| 930 |
+
# When we created the data, we kept track of the alignment between original
|
| 931 |
+
# (whitespace tokenized) tokens and our WordPiece tokenized tokens. So
|
| 932 |
+
# now `orig_text` contains the span of our original text corresponding to the
|
| 933 |
+
# span that we predicted.
|
| 934 |
+
#
|
| 935 |
+
# However, `orig_text` may contain extra characters that we don't want in
|
| 936 |
+
# our prediction.
|
| 937 |
+
#
|
| 938 |
+
# For example, let's say:
|
| 939 |
+
# pred_text = steve smith
|
| 940 |
+
# orig_text = Steve Smith's
|
| 941 |
+
#
|
| 942 |
+
# We don't want to return `orig_text` because it contains the extra "'s".
|
| 943 |
+
#
|
| 944 |
+
# We don't want to return `pred_text` because it's already been normalized
|
| 945 |
+
# (the SQuAD eval script also does punctuation stripping/lower casing but
|
| 946 |
+
# our tokenizer does additional normalization like stripping accent
|
| 947 |
+
# characters).
|
| 948 |
+
#
|
| 949 |
+
# What we really want to return is "Steve Smith".
|
| 950 |
+
#
|
| 951 |
+
# Therefore, we have to apply a semi-complicated alignment heruistic between
|
| 952 |
+
# `pred_text` and `orig_text` to get a character-to-charcter alignment. This
|
| 953 |
+
# can fail in certain cases in which case we just return `orig_text`.
|
| 954 |
+
|
| 955 |
+
def _strip_spaces(text):
|
| 956 |
+
ns_chars = []
|
| 957 |
+
ns_to_s_map = collections.OrderedDict()
|
| 958 |
+
for (i, c) in enumerate(text):
|
| 959 |
+
if c == " ":
|
| 960 |
+
continue
|
| 961 |
+
ns_to_s_map[len(ns_chars)] = i
|
| 962 |
+
ns_chars.append(c)
|
| 963 |
+
ns_text = "".join(ns_chars)
|
| 964 |
+
return (ns_text, ns_to_s_map)
|
| 965 |
+
|
| 966 |
+
# We first tokenize `orig_text`, strip whitespace from the result
|
| 967 |
+
# and `pred_text`, and check if they are the same length. If they are
|
| 968 |
+
# NOT the same length, the heuristic has failed. If they are the same
|
| 969 |
+
# length, we assume the characters are one-to-one aligned.
|
| 970 |
+
tokenizer = tokenization.BasicTokenizer(do_lower_case=do_lower_case)
|
| 971 |
+
|
| 972 |
+
tok_text = " ".join(tokenizer.tokenize(orig_text))
|
| 973 |
+
|
| 974 |
+
start_position = tok_text.find(pred_text)
|
| 975 |
+
if start_position == -1:
|
| 976 |
+
if FLAGS.verbose_logging:
|
| 977 |
+
tf.logging.info(
|
| 978 |
+
"Unable to find text: '%s' in '%s'" % (pred_text, orig_text))
|
| 979 |
+
return orig_text
|
| 980 |
+
end_position = start_position + len(pred_text) - 1
|
| 981 |
+
|
| 982 |
+
(orig_ns_text, orig_ns_to_s_map) = _strip_spaces(orig_text)
|
| 983 |
+
(tok_ns_text, tok_ns_to_s_map) = _strip_spaces(tok_text)
|
| 984 |
+
|
| 985 |
+
if len(orig_ns_text) != len(tok_ns_text):
|
| 986 |
+
if FLAGS.verbose_logging:
|
| 987 |
+
tf.logging.info("Length not equal after stripping spaces: '%s' vs '%s'",
|
| 988 |
+
orig_ns_text, tok_ns_text)
|
| 989 |
+
return orig_text
|
| 990 |
+
|
| 991 |
+
# We then project the characters in `pred_text` back to `orig_text` using
|
| 992 |
+
# the character-to-character alignment.
|
| 993 |
+
tok_s_to_ns_map = {}
|
| 994 |
+
for (i, tok_index) in six.iteritems(tok_ns_to_s_map):
|
| 995 |
+
tok_s_to_ns_map[tok_index] = i
|
| 996 |
+
|
| 997 |
+
orig_start_position = None
|
| 998 |
+
if start_position in tok_s_to_ns_map:
|
| 999 |
+
ns_start_position = tok_s_to_ns_map[start_position]
|
| 1000 |
+
if ns_start_position in orig_ns_to_s_map:
|
| 1001 |
+
orig_start_position = orig_ns_to_s_map[ns_start_position]
|
| 1002 |
+
|
| 1003 |
+
if orig_start_position is None:
|
| 1004 |
+
if FLAGS.verbose_logging:
|
| 1005 |
+
tf.logging.info("Couldn't map start position")
|
| 1006 |
+
return orig_text
|
| 1007 |
+
|
| 1008 |
+
orig_end_position = None
|
| 1009 |
+
if end_position in tok_s_to_ns_map:
|
| 1010 |
+
ns_end_position = tok_s_to_ns_map[end_position]
|
| 1011 |
+
if ns_end_position in orig_ns_to_s_map:
|
| 1012 |
+
orig_end_position = orig_ns_to_s_map[ns_end_position]
|
| 1013 |
+
|
| 1014 |
+
if orig_end_position is None:
|
| 1015 |
+
if FLAGS.verbose_logging:
|
| 1016 |
+
tf.logging.info("Couldn't map end position")
|
| 1017 |
+
return orig_text
|
| 1018 |
+
|
| 1019 |
+
output_text = orig_text[orig_start_position:(orig_end_position + 1)]
|
| 1020 |
+
return output_text
|
| 1021 |
+
|
| 1022 |
+
|
| 1023 |
+
def _get_best_indexes(logits, n_best_size):
|
| 1024 |
+
"""Get the n-best logits from a list."""
|
| 1025 |
+
index_and_score = sorted(enumerate(logits), key=lambda x: x[1], reverse=True)
|
| 1026 |
+
|
| 1027 |
+
best_indexes = []
|
| 1028 |
+
for i in range(len(index_and_score)):
|
| 1029 |
+
if i >= n_best_size:
|
| 1030 |
+
break
|
| 1031 |
+
best_indexes.append(index_and_score[i][0])
|
| 1032 |
+
return best_indexes
|
| 1033 |
+
|
| 1034 |
+
|
| 1035 |
+
def _compute_softmax(scores):
|
| 1036 |
+
"""Compute softmax probability over raw logits."""
|
| 1037 |
+
if not scores:
|
| 1038 |
+
return []
|
| 1039 |
+
|
| 1040 |
+
max_score = None
|
| 1041 |
+
for score in scores:
|
| 1042 |
+
if max_score is None or score > max_score:
|
| 1043 |
+
max_score = score
|
| 1044 |
+
|
| 1045 |
+
exp_scores = []
|
| 1046 |
+
total_sum = 0.0
|
| 1047 |
+
for score in scores:
|
| 1048 |
+
x = math.exp(score - max_score)
|
| 1049 |
+
exp_scores.append(x)
|
| 1050 |
+
total_sum += x
|
| 1051 |
+
|
| 1052 |
+
probs = []
|
| 1053 |
+
for score in exp_scores:
|
| 1054 |
+
probs.append(score / total_sum)
|
| 1055 |
+
return probs
|
| 1056 |
+
|
| 1057 |
+
|
| 1058 |
+
class FeatureWriter(object):
|
| 1059 |
+
"""Writes InputFeature to TF example file."""
|
| 1060 |
+
|
| 1061 |
+
def __init__(self, filename, is_training):
|
| 1062 |
+
self.filename = filename
|
| 1063 |
+
self.is_training = is_training
|
| 1064 |
+
self.num_features = 0
|
| 1065 |
+
self._writer = tf.python_io.TFRecordWriter(filename)
|
| 1066 |
+
|
| 1067 |
+
def process_feature(self, feature):
|
| 1068 |
+
"""Write a InputFeature to the TFRecordWriter as a tf.train.Example."""
|
| 1069 |
+
self.num_features += 1
|
| 1070 |
+
|
| 1071 |
+
def create_int_feature(values):
|
| 1072 |
+
feature = tf.train.Feature(
|
| 1073 |
+
int64_list=tf.train.Int64List(value=list(values)))
|
| 1074 |
+
return feature
|
| 1075 |
+
|
| 1076 |
+
features = collections.OrderedDict()
|
| 1077 |
+
features["unique_ids"] = create_int_feature([feature.unique_id])
|
| 1078 |
+
features["input_ids"] = create_int_feature(feature.input_ids)
|
| 1079 |
+
features["input_mask"] = create_int_feature(feature.input_mask)
|
| 1080 |
+
features["segment_ids"] = create_int_feature(feature.segment_ids)
|
| 1081 |
+
|
| 1082 |
+
if self.is_training:
|
| 1083 |
+
features["start_positions"] = create_int_feature([feature.start_position])
|
| 1084 |
+
features["end_positions"] = create_int_feature([feature.end_position])
|
| 1085 |
+
impossible = 0
|
| 1086 |
+
if feature.is_impossible:
|
| 1087 |
+
impossible = 1
|
| 1088 |
+
features["is_impossible"] = create_int_feature([impossible])
|
| 1089 |
+
|
| 1090 |
+
tf_example = tf.train.Example(features=tf.train.Features(feature=features))
|
| 1091 |
+
self._writer.write(tf_example.SerializeToString())
|
| 1092 |
+
|
| 1093 |
+
def close(self):
|
| 1094 |
+
self._writer.close()
|
| 1095 |
+
|
| 1096 |
+
|
| 1097 |
+
def validate_flags_or_throw(bert_config):
|
| 1098 |
+
"""Validate the input FLAGS or throw an exception."""
|
| 1099 |
+
tokenization.validate_case_matches_checkpoint(FLAGS.do_lower_case,
|
| 1100 |
+
FLAGS.init_checkpoint)
|
| 1101 |
+
|
| 1102 |
+
if not FLAGS.do_train and not FLAGS.do_predict:
|
| 1103 |
+
raise ValueError("At least one of `do_train` or `do_predict` must be True.")
|
| 1104 |
+
|
| 1105 |
+
if FLAGS.do_train:
|
| 1106 |
+
if not FLAGS.train_file:
|
| 1107 |
+
raise ValueError(
|
| 1108 |
+
"If `do_train` is True, then `train_file` must be specified.")
|
| 1109 |
+
if FLAGS.do_predict:
|
| 1110 |
+
if not FLAGS.predict_file:
|
| 1111 |
+
raise ValueError(
|
| 1112 |
+
"If `do_predict` is True, then `predict_file` must be specified.")
|
| 1113 |
+
|
| 1114 |
+
if FLAGS.max_seq_length > bert_config.max_position_embeddings:
|
| 1115 |
+
raise ValueError(
|
| 1116 |
+
"Cannot use sequence length %d because the BERT model "
|
| 1117 |
+
"was only trained up to sequence length %d" %
|
| 1118 |
+
(FLAGS.max_seq_length, bert_config.max_position_embeddings))
|
| 1119 |
+
|
| 1120 |
+
if FLAGS.max_seq_length <= FLAGS.max_query_length + 3:
|
| 1121 |
+
raise ValueError(
|
| 1122 |
+
"The max_seq_length (%d) must be greater than max_query_length "
|
| 1123 |
+
"(%d) + 3" % (FLAGS.max_seq_length, FLAGS.max_query_length))
|
| 1124 |
+
|
| 1125 |
+
|
| 1126 |
+
def main(_):
|
| 1127 |
+
tf.logging.set_verbosity(tf.logging.INFO)
|
| 1128 |
+
|
| 1129 |
+
bert_config = modeling.BertConfig.from_json_file(FLAGS.bert_config_file)
|
| 1130 |
+
|
| 1131 |
+
validate_flags_or_throw(bert_config)
|
| 1132 |
+
|
| 1133 |
+
tf.gfile.MakeDirs(FLAGS.output_dir)
|
| 1134 |
+
|
| 1135 |
+
tokenizer = tokenization.FullTokenizer(
|
| 1136 |
+
vocab_file=FLAGS.vocab_file, do_lower_case=FLAGS.do_lower_case)
|
| 1137 |
+
|
| 1138 |
+
tpu_cluster_resolver = None
|
| 1139 |
+
if FLAGS.use_tpu and FLAGS.tpu_name:
|
| 1140 |
+
tpu_cluster_resolver = tf.contrib.cluster_resolver.TPUClusterResolver(
|
| 1141 |
+
FLAGS.tpu_name, zone=FLAGS.tpu_zone, project=FLAGS.gcp_project)
|
| 1142 |
+
|
| 1143 |
+
is_per_host = tf.contrib.tpu.InputPipelineConfig.PER_HOST_V2
|
| 1144 |
+
run_config = tf.contrib.tpu.RunConfig(
|
| 1145 |
+
cluster=tpu_cluster_resolver,
|
| 1146 |
+
master=FLAGS.master,
|
| 1147 |
+
model_dir=FLAGS.output_dir,
|
| 1148 |
+
save_checkpoints_steps=FLAGS.save_checkpoints_steps,
|
| 1149 |
+
tpu_config=tf.contrib.tpu.TPUConfig(
|
| 1150 |
+
iterations_per_loop=FLAGS.iterations_per_loop,
|
| 1151 |
+
num_shards=FLAGS.num_tpu_cores,
|
| 1152 |
+
per_host_input_for_training=is_per_host))
|
| 1153 |
+
|
| 1154 |
+
train_examples = None
|
| 1155 |
+
num_train_steps = None
|
| 1156 |
+
num_warmup_steps = None
|
| 1157 |
+
if FLAGS.do_train:
|
| 1158 |
+
train_examples = read_squad_examples(
|
| 1159 |
+
input_file=FLAGS.train_file, is_training=True)
|
| 1160 |
+
num_train_steps = int(
|
| 1161 |
+
len(train_examples) / FLAGS.train_batch_size * FLAGS.num_train_epochs)
|
| 1162 |
+
num_warmup_steps = int(num_train_steps * FLAGS.warmup_proportion)
|
| 1163 |
+
|
| 1164 |
+
# Pre-shuffle the input to avoid having to make a very large shuffle
|
| 1165 |
+
# buffer in in the `input_fn`.
|
| 1166 |
+
rng = random.Random(12345)
|
| 1167 |
+
rng.shuffle(train_examples)
|
| 1168 |
+
|
| 1169 |
+
model_fn = model_fn_builder(
|
| 1170 |
+
bert_config=bert_config,
|
| 1171 |
+
init_checkpoint=FLAGS.init_checkpoint,
|
| 1172 |
+
learning_rate=FLAGS.learning_rate,
|
| 1173 |
+
num_train_steps=num_train_steps,
|
| 1174 |
+
num_warmup_steps=num_warmup_steps,
|
| 1175 |
+
use_tpu=FLAGS.use_tpu,
|
| 1176 |
+
use_one_hot_embeddings=FLAGS.use_tpu)
|
| 1177 |
+
|
| 1178 |
+
# If TPU is not available, this will fall back to normal Estimator on CPU
|
| 1179 |
+
# or GPU.
|
| 1180 |
+
estimator = tf.contrib.tpu.TPUEstimator(
|
| 1181 |
+
use_tpu=FLAGS.use_tpu,
|
| 1182 |
+
model_fn=model_fn,
|
| 1183 |
+
config=run_config,
|
| 1184 |
+
train_batch_size=FLAGS.train_batch_size,
|
| 1185 |
+
predict_batch_size=FLAGS.predict_batch_size)
|
| 1186 |
+
|
| 1187 |
+
if FLAGS.do_train:
|
| 1188 |
+
# We write to a temporary file to avoid storing very large constant tensors
|
| 1189 |
+
# in memory.
|
| 1190 |
+
train_writer = FeatureWriter(
|
| 1191 |
+
filename=os.path.join(FLAGS.output_dir, "train.tf_record"),
|
| 1192 |
+
is_training=True)
|
| 1193 |
+
convert_examples_to_features(
|
| 1194 |
+
examples=train_examples,
|
| 1195 |
+
tokenizer=tokenizer,
|
| 1196 |
+
max_seq_length=FLAGS.max_seq_length,
|
| 1197 |
+
doc_stride=FLAGS.doc_stride,
|
| 1198 |
+
max_query_length=FLAGS.max_query_length,
|
| 1199 |
+
is_training=True,
|
| 1200 |
+
output_fn=train_writer.process_feature)
|
| 1201 |
+
train_writer.close()
|
| 1202 |
+
|
| 1203 |
+
tf.logging.info("***** Running training *****")
|
| 1204 |
+
tf.logging.info(" Num orig examples = %d", len(train_examples))
|
| 1205 |
+
tf.logging.info(" Num split examples = %d", train_writer.num_features)
|
| 1206 |
+
tf.logging.info(" Batch size = %d", FLAGS.train_batch_size)
|
| 1207 |
+
tf.logging.info(" Num steps = %d", num_train_steps)
|
| 1208 |
+
del train_examples
|
| 1209 |
+
|
| 1210 |
+
train_input_fn = input_fn_builder(
|
| 1211 |
+
input_file=train_writer.filename,
|
| 1212 |
+
seq_length=FLAGS.max_seq_length,
|
| 1213 |
+
is_training=True,
|
| 1214 |
+
drop_remainder=True)
|
| 1215 |
+
estimator.train(input_fn=train_input_fn, max_steps=num_train_steps)
|
| 1216 |
+
|
| 1217 |
+
if FLAGS.do_predict:
|
| 1218 |
+
eval_examples = read_squad_examples(
|
| 1219 |
+
input_file=FLAGS.predict_file, is_training=False)
|
| 1220 |
+
|
| 1221 |
+
eval_writer = FeatureWriter(
|
| 1222 |
+
filename=os.path.join(FLAGS.output_dir, "eval.tf_record"),
|
| 1223 |
+
is_training=False)
|
| 1224 |
+
eval_features = []
|
| 1225 |
+
|
| 1226 |
+
def append_feature(feature):
|
| 1227 |
+
eval_features.append(feature)
|
| 1228 |
+
eval_writer.process_feature(feature)
|
| 1229 |
+
|
| 1230 |
+
convert_examples_to_features(
|
| 1231 |
+
examples=eval_examples,
|
| 1232 |
+
tokenizer=tokenizer,
|
| 1233 |
+
max_seq_length=FLAGS.max_seq_length,
|
| 1234 |
+
doc_stride=FLAGS.doc_stride,
|
| 1235 |
+
max_query_length=FLAGS.max_query_length,
|
| 1236 |
+
is_training=False,
|
| 1237 |
+
output_fn=append_feature)
|
| 1238 |
+
eval_writer.close()
|
| 1239 |
+
|
| 1240 |
+
tf.logging.info("***** Running predictions *****")
|
| 1241 |
+
tf.logging.info(" Num orig examples = %d", len(eval_examples))
|
| 1242 |
+
tf.logging.info(" Num split examples = %d", len(eval_features))
|
| 1243 |
+
tf.logging.info(" Batch size = %d", FLAGS.predict_batch_size)
|
| 1244 |
+
|
| 1245 |
+
all_results = []
|
| 1246 |
+
|
| 1247 |
+
predict_input_fn = input_fn_builder(
|
| 1248 |
+
input_file=eval_writer.filename,
|
| 1249 |
+
seq_length=FLAGS.max_seq_length,
|
| 1250 |
+
is_training=False,
|
| 1251 |
+
drop_remainder=False)
|
| 1252 |
+
|
| 1253 |
+
# If running eval on the TPU, you will need to specify the number of
|
| 1254 |
+
# steps.
|
| 1255 |
+
all_results = []
|
| 1256 |
+
for result in estimator.predict(
|
| 1257 |
+
predict_input_fn, yield_single_examples=True):
|
| 1258 |
+
if len(all_results) % 1000 == 0:
|
| 1259 |
+
tf.logging.info("Processing example: %d" % (len(all_results)))
|
| 1260 |
+
unique_id = int(result["unique_ids"])
|
| 1261 |
+
start_logits = [float(x) for x in result["start_logits"].flat]
|
| 1262 |
+
end_logits = [float(x) for x in result["end_logits"].flat]
|
| 1263 |
+
all_results.append(
|
| 1264 |
+
RawResult(
|
| 1265 |
+
unique_id=unique_id,
|
| 1266 |
+
start_logits=start_logits,
|
| 1267 |
+
end_logits=end_logits))
|
| 1268 |
+
|
| 1269 |
+
output_prediction_file = os.path.join(FLAGS.output_dir, "predictions.json")
|
| 1270 |
+
output_nbest_file = os.path.join(FLAGS.output_dir, "nbest_predictions.json")
|
| 1271 |
+
output_null_log_odds_file = os.path.join(FLAGS.output_dir, "null_odds.json")
|
| 1272 |
+
|
| 1273 |
+
write_predictions(eval_examples, eval_features, all_results,
|
| 1274 |
+
FLAGS.n_best_size, FLAGS.max_answer_length,
|
| 1275 |
+
FLAGS.do_lower_case, output_prediction_file,
|
| 1276 |
+
output_nbest_file, output_null_log_odds_file)
|
| 1277 |
+
|
| 1278 |
+
|
| 1279 |
+
if __name__ == "__main__":
|
| 1280 |
+
flags.mark_flag_as_required("vocab_file")
|
| 1281 |
+
flags.mark_flag_as_required("bert_config_file")
|
| 1282 |
+
flags.mark_flag_as_required("output_dir")
|
| 1283 |
+
tf.app.run()
|
bert-master/bert-master/sample_text.txt
ADDED
|
@@ -0,0 +1,33 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
This text is included to make sure Unicode is handled properly: 力加勝北区ᴵᴺᵀᵃছজটডণত
|
| 2 |
+
Text should be one-sentence-per-line, with empty lines between documents.
|
| 3 |
+
This sample text is public domain and was randomly selected from Project Guttenberg.
|
| 4 |
+
|
| 5 |
+
The rain had only ceased with the gray streaks of morning at Blazing Star, and the settlement awoke to a moral sense of cleanliness, and the finding of forgotten knives, tin cups, and smaller camp utensils, where the heavy showers had washed away the debris and dust heaps before the cabin doors.
|
| 6 |
+
Indeed, it was recorded in Blazing Star that a fortunate early riser had once picked up on the highway a solid chunk of gold quartz which the rain had freed from its incumbering soil, and washed into immediate and glittering popularity.
|
| 7 |
+
Possibly this may have been the reason why early risers in that locality, during the rainy season, adopted a thoughtful habit of body, and seldom lifted their eyes to the rifted or india-ink washed skies above them.
|
| 8 |
+
"Cass" Beard had risen early that morning, but not with a view to discovery.
|
| 9 |
+
A leak in his cabin roof,--quite consistent with his careless, improvident habits,--had roused him at 4 A. M., with a flooded "bunk" and wet blankets.
|
| 10 |
+
The chips from his wood pile refused to kindle a fire to dry his bed-clothes, and he had recourse to a more provident neighbor's to supply the deficiency.
|
| 11 |
+
This was nearly opposite.
|
| 12 |
+
Mr. Cassius crossed the highway, and stopped suddenly.
|
| 13 |
+
Something glittered in the nearest red pool before him.
|
| 14 |
+
Gold, surely!
|
| 15 |
+
But, wonderful to relate, not an irregular, shapeless fragment of crude ore, fresh from Nature's crucible, but a bit of jeweler's handicraft in the form of a plain gold ring.
|
| 16 |
+
Looking at it more attentively, he saw that it bore the inscription, "May to Cass."
|
| 17 |
+
Like most of his fellow gold-seekers, Cass was superstitious.
|
| 18 |
+
|
| 19 |
+
The fountain of classic wisdom, Hypatia herself.
|
| 20 |
+
As the ancient sage--the name is unimportant to a monk--pumped water nightly that he might study by day, so I, the guardian of cloaks and parasols, at the sacred doors of her lecture-room, imbibe celestial knowledge.
|
| 21 |
+
From my youth I felt in me a soul above the matter-entangled herd.
|
| 22 |
+
She revealed to me the glorious fact, that I am a spark of Divinity itself.
|
| 23 |
+
A fallen star, I am, sir!' continued he, pensively, stroking his lean stomach--'a fallen star!--fallen, if the dignity of philosophy will allow of the simile, among the hogs of the lower world--indeed, even into the hog-bucket itself. Well, after all, I will show you the way to the Archbishop's.
|
| 24 |
+
There is a philosophic pleasure in opening one's treasures to the modest young.
|
| 25 |
+
Perhaps you will assist me by carrying this basket of fruit?' And the little man jumped up, put his basket on Philammon's head, and trotted off up a neighbouring street.
|
| 26 |
+
Philammon followed, half contemptuous, half wondering at what this philosophy might be, which could feed the self-conceit of anything so abject as his ragged little apish guide;
|
| 27 |
+
but the novel roar and whirl of the street, the perpetual stream of busy faces, the line of curricles, palanquins, laden asses, camels, elephants, which met and passed him, and squeezed him up steps and into doorways, as they threaded their way through the great Moon-gate into the ample street beyond, drove everything from his mind but wondering curiosity, and a vague, helpless dread of that great living wilderness, more terrible than any dead wilderness of sand which he had left behind.
|
| 28 |
+
Already he longed for the repose, the silence of the Laura--for faces which knew him and smiled upon him; but it was too late to turn back now.
|
| 29 |
+
His guide held on for more than a mile up the great main street, crossed in the centre of the city, at right angles, by one equally magnificent, at each end of which, miles away, appeared, dim and distant over the heads of the living stream of passengers, the yellow sand-hills of the desert;
|
| 30 |
+
while at the end of the vista in front of them gleamed the blue harbour, through a network of countless masts.
|
| 31 |
+
At last they reached the quay at the opposite end of the street;
|
| 32 |
+
and there burst on Philammon's astonished eyes a vast semicircle of blue sea, ringed with palaces and towers.
|
| 33 |
+
He stopped involuntarily; and his little guide stopped also, and looked askance at the young monk, to watch the effect which that grand panorama should produce on him.
|
bert-master/bert-master/tokenization.py
ADDED
|
@@ -0,0 +1,399 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# coding=utf-8
|
| 2 |
+
# Copyright 2018 The Google AI Language Team Authors.
|
| 3 |
+
#
|
| 4 |
+
# Licensed under the Apache License, Version 2.0 (the "License");
|
| 5 |
+
# you may not use this file except in compliance with the License.
|
| 6 |
+
# You may obtain a copy of the License at
|
| 7 |
+
#
|
| 8 |
+
# http://www.apache.org/licenses/LICENSE-2.0
|
| 9 |
+
#
|
| 10 |
+
# Unless required by applicable law or agreed to in writing, software
|
| 11 |
+
# distributed under the License is distributed on an "AS IS" BASIS,
|
| 12 |
+
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
| 13 |
+
# See the License for the specific language governing permissions and
|
| 14 |
+
# limitations under the License.
|
| 15 |
+
"""Tokenization classes."""
|
| 16 |
+
|
| 17 |
+
from __future__ import absolute_import
|
| 18 |
+
from __future__ import division
|
| 19 |
+
from __future__ import print_function
|
| 20 |
+
|
| 21 |
+
import collections
|
| 22 |
+
import re
|
| 23 |
+
import unicodedata
|
| 24 |
+
import six
|
| 25 |
+
import tensorflow as tf
|
| 26 |
+
|
| 27 |
+
|
| 28 |
+
def validate_case_matches_checkpoint(do_lower_case, init_checkpoint):
|
| 29 |
+
"""Checks whether the casing config is consistent with the checkpoint name."""
|
| 30 |
+
|
| 31 |
+
# The casing has to be passed in by the user and there is no explicit check
|
| 32 |
+
# as to whether it matches the checkpoint. The casing information probably
|
| 33 |
+
# should have been stored in the bert_config.json file, but it's not, so
|
| 34 |
+
# we have to heuristically detect it to validate.
|
| 35 |
+
|
| 36 |
+
if not init_checkpoint:
|
| 37 |
+
return
|
| 38 |
+
|
| 39 |
+
m = re.match("^.*?([A-Za-z0-9_-]+)/bert_model.ckpt", init_checkpoint)
|
| 40 |
+
if m is None:
|
| 41 |
+
return
|
| 42 |
+
|
| 43 |
+
model_name = m.group(1)
|
| 44 |
+
|
| 45 |
+
lower_models = [
|
| 46 |
+
"uncased_L-24_H-1024_A-16", "uncased_L-12_H-768_A-12",
|
| 47 |
+
"multilingual_L-12_H-768_A-12", "chinese_L-12_H-768_A-12"
|
| 48 |
+
]
|
| 49 |
+
|
| 50 |
+
cased_models = [
|
| 51 |
+
"cased_L-12_H-768_A-12", "cased_L-24_H-1024_A-16",
|
| 52 |
+
"multi_cased_L-12_H-768_A-12"
|
| 53 |
+
]
|
| 54 |
+
|
| 55 |
+
is_bad_config = False
|
| 56 |
+
if model_name in lower_models and not do_lower_case:
|
| 57 |
+
is_bad_config = True
|
| 58 |
+
actual_flag = "False"
|
| 59 |
+
case_name = "lowercased"
|
| 60 |
+
opposite_flag = "True"
|
| 61 |
+
|
| 62 |
+
if model_name in cased_models and do_lower_case:
|
| 63 |
+
is_bad_config = True
|
| 64 |
+
actual_flag = "True"
|
| 65 |
+
case_name = "cased"
|
| 66 |
+
opposite_flag = "False"
|
| 67 |
+
|
| 68 |
+
if is_bad_config:
|
| 69 |
+
raise ValueError(
|
| 70 |
+
"You passed in `--do_lower_case=%s` with `--init_checkpoint=%s`. "
|
| 71 |
+
"However, `%s` seems to be a %s model, so you "
|
| 72 |
+
"should pass in `--do_lower_case=%s` so that the fine-tuning matches "
|
| 73 |
+
"how the model was pre-training. If this error is wrong, please "
|
| 74 |
+
"just comment out this check." % (actual_flag, init_checkpoint,
|
| 75 |
+
model_name, case_name, opposite_flag))
|
| 76 |
+
|
| 77 |
+
|
| 78 |
+
def convert_to_unicode(text):
|
| 79 |
+
"""Converts `text` to Unicode (if it's not already), assuming utf-8 input."""
|
| 80 |
+
if six.PY3:
|
| 81 |
+
if isinstance(text, str):
|
| 82 |
+
return text
|
| 83 |
+
elif isinstance(text, bytes):
|
| 84 |
+
return text.decode("utf-8", "ignore")
|
| 85 |
+
else:
|
| 86 |
+
raise ValueError("Unsupported string type: %s" % (type(text)))
|
| 87 |
+
elif six.PY2:
|
| 88 |
+
if isinstance(text, str):
|
| 89 |
+
return text.decode("utf-8", "ignore")
|
| 90 |
+
elif isinstance(text, unicode):
|
| 91 |
+
return text
|
| 92 |
+
else:
|
| 93 |
+
raise ValueError("Unsupported string type: %s" % (type(text)))
|
| 94 |
+
else:
|
| 95 |
+
raise ValueError("Not running on Python2 or Python 3?")
|
| 96 |
+
|
| 97 |
+
|
| 98 |
+
def printable_text(text):
|
| 99 |
+
"""Returns text encoded in a way suitable for print or `tf.logging`."""
|
| 100 |
+
|
| 101 |
+
# These functions want `str` for both Python2 and Python3, but in one case
|
| 102 |
+
# it's a Unicode string and in the other it's a byte string.
|
| 103 |
+
if six.PY3:
|
| 104 |
+
if isinstance(text, str):
|
| 105 |
+
return text
|
| 106 |
+
elif isinstance(text, bytes):
|
| 107 |
+
return text.decode("utf-8", "ignore")
|
| 108 |
+
else:
|
| 109 |
+
raise ValueError("Unsupported string type: %s" % (type(text)))
|
| 110 |
+
elif six.PY2:
|
| 111 |
+
if isinstance(text, str):
|
| 112 |
+
return text
|
| 113 |
+
elif isinstance(text, unicode):
|
| 114 |
+
return text.encode("utf-8")
|
| 115 |
+
else:
|
| 116 |
+
raise ValueError("Unsupported string type: %s" % (type(text)))
|
| 117 |
+
else:
|
| 118 |
+
raise ValueError("Not running on Python2 or Python 3?")
|
| 119 |
+
|
| 120 |
+
|
| 121 |
+
def load_vocab(vocab_file):
|
| 122 |
+
"""Loads a vocabulary file into a dictionary."""
|
| 123 |
+
vocab = collections.OrderedDict()
|
| 124 |
+
index = 0
|
| 125 |
+
with tf.gfile.GFile(vocab_file, "r") as reader:
|
| 126 |
+
while True:
|
| 127 |
+
token = convert_to_unicode(reader.readline())
|
| 128 |
+
if not token:
|
| 129 |
+
break
|
| 130 |
+
token = token.strip()
|
| 131 |
+
vocab[token] = index
|
| 132 |
+
index += 1
|
| 133 |
+
return vocab
|
| 134 |
+
|
| 135 |
+
|
| 136 |
+
def convert_by_vocab(vocab, items):
|
| 137 |
+
"""Converts a sequence of [tokens|ids] using the vocab."""
|
| 138 |
+
output = []
|
| 139 |
+
for item in items:
|
| 140 |
+
output.append(vocab[item])
|
| 141 |
+
return output
|
| 142 |
+
|
| 143 |
+
|
| 144 |
+
def convert_tokens_to_ids(vocab, tokens):
|
| 145 |
+
return convert_by_vocab(vocab, tokens)
|
| 146 |
+
|
| 147 |
+
|
| 148 |
+
def convert_ids_to_tokens(inv_vocab, ids):
|
| 149 |
+
return convert_by_vocab(inv_vocab, ids)
|
| 150 |
+
|
| 151 |
+
|
| 152 |
+
def whitespace_tokenize(text):
|
| 153 |
+
"""Runs basic whitespace cleaning and splitting on a piece of text."""
|
| 154 |
+
text = text.strip()
|
| 155 |
+
if not text:
|
| 156 |
+
return []
|
| 157 |
+
tokens = text.split()
|
| 158 |
+
return tokens
|
| 159 |
+
|
| 160 |
+
|
| 161 |
+
class FullTokenizer(object):
|
| 162 |
+
"""Runs end-to-end tokenziation."""
|
| 163 |
+
|
| 164 |
+
def __init__(self, vocab_file, do_lower_case=True):
|
| 165 |
+
self.vocab = load_vocab(vocab_file)
|
| 166 |
+
self.inv_vocab = {v: k for k, v in self.vocab.items()}
|
| 167 |
+
self.basic_tokenizer = BasicTokenizer(do_lower_case=do_lower_case)
|
| 168 |
+
self.wordpiece_tokenizer = WordpieceTokenizer(vocab=self.vocab)
|
| 169 |
+
|
| 170 |
+
def tokenize(self, text):
|
| 171 |
+
split_tokens = []
|
| 172 |
+
for token in self.basic_tokenizer.tokenize(text):
|
| 173 |
+
for sub_token in self.wordpiece_tokenizer.tokenize(token):
|
| 174 |
+
split_tokens.append(sub_token)
|
| 175 |
+
|
| 176 |
+
return split_tokens
|
| 177 |
+
|
| 178 |
+
def convert_tokens_to_ids(self, tokens):
|
| 179 |
+
return convert_by_vocab(self.vocab, tokens)
|
| 180 |
+
|
| 181 |
+
def convert_ids_to_tokens(self, ids):
|
| 182 |
+
return convert_by_vocab(self.inv_vocab, ids)
|
| 183 |
+
|
| 184 |
+
|
| 185 |
+
class BasicTokenizer(object):
|
| 186 |
+
"""Runs basic tokenization (punctuation splitting, lower casing, etc.)."""
|
| 187 |
+
|
| 188 |
+
def __init__(self, do_lower_case=True):
|
| 189 |
+
"""Constructs a BasicTokenizer.
|
| 190 |
+
|
| 191 |
+
Args:
|
| 192 |
+
do_lower_case: Whether to lower case the input.
|
| 193 |
+
"""
|
| 194 |
+
self.do_lower_case = do_lower_case
|
| 195 |
+
|
| 196 |
+
def tokenize(self, text):
|
| 197 |
+
"""Tokenizes a piece of text."""
|
| 198 |
+
text = convert_to_unicode(text)
|
| 199 |
+
text = self._clean_text(text)
|
| 200 |
+
|
| 201 |
+
# This was added on November 1st, 2018 for the multilingual and Chinese
|
| 202 |
+
# models. This is also applied to the English models now, but it doesn't
|
| 203 |
+
# matter since the English models were not trained on any Chinese data
|
| 204 |
+
# and generally don't have any Chinese data in them (there are Chinese
|
| 205 |
+
# characters in the vocabulary because Wikipedia does have some Chinese
|
| 206 |
+
# words in the English Wikipedia.).
|
| 207 |
+
text = self._tokenize_chinese_chars(text)
|
| 208 |
+
|
| 209 |
+
orig_tokens = whitespace_tokenize(text)
|
| 210 |
+
split_tokens = []
|
| 211 |
+
for token in orig_tokens:
|
| 212 |
+
if self.do_lower_case:
|
| 213 |
+
token = token.lower()
|
| 214 |
+
token = self._run_strip_accents(token)
|
| 215 |
+
split_tokens.extend(self._run_split_on_punc(token))
|
| 216 |
+
|
| 217 |
+
output_tokens = whitespace_tokenize(" ".join(split_tokens))
|
| 218 |
+
return output_tokens
|
| 219 |
+
|
| 220 |
+
def _run_strip_accents(self, text):
|
| 221 |
+
"""Strips accents from a piece of text."""
|
| 222 |
+
text = unicodedata.normalize("NFD", text)
|
| 223 |
+
output = []
|
| 224 |
+
for char in text:
|
| 225 |
+
cat = unicodedata.category(char)
|
| 226 |
+
if cat == "Mn":
|
| 227 |
+
continue
|
| 228 |
+
output.append(char)
|
| 229 |
+
return "".join(output)
|
| 230 |
+
|
| 231 |
+
def _run_split_on_punc(self, text):
|
| 232 |
+
"""Splits punctuation on a piece of text."""
|
| 233 |
+
chars = list(text)
|
| 234 |
+
i = 0
|
| 235 |
+
start_new_word = True
|
| 236 |
+
output = []
|
| 237 |
+
while i < len(chars):
|
| 238 |
+
char = chars[i]
|
| 239 |
+
if _is_punctuation(char):
|
| 240 |
+
output.append([char])
|
| 241 |
+
start_new_word = True
|
| 242 |
+
else:
|
| 243 |
+
if start_new_word:
|
| 244 |
+
output.append([])
|
| 245 |
+
start_new_word = False
|
| 246 |
+
output[-1].append(char)
|
| 247 |
+
i += 1
|
| 248 |
+
|
| 249 |
+
return ["".join(x) for x in output]
|
| 250 |
+
|
| 251 |
+
def _tokenize_chinese_chars(self, text):
|
| 252 |
+
"""Adds whitespace around any CJK character."""
|
| 253 |
+
output = []
|
| 254 |
+
for char in text:
|
| 255 |
+
cp = ord(char)
|
| 256 |
+
if self._is_chinese_char(cp):
|
| 257 |
+
output.append(" ")
|
| 258 |
+
output.append(char)
|
| 259 |
+
output.append(" ")
|
| 260 |
+
else:
|
| 261 |
+
output.append(char)
|
| 262 |
+
return "".join(output)
|
| 263 |
+
|
| 264 |
+
def _is_chinese_char(self, cp):
|
| 265 |
+
"""Checks whether CP is the codepoint of a CJK character."""
|
| 266 |
+
# This defines a "chinese character" as anything in the CJK Unicode block:
|
| 267 |
+
# https://en.wikipedia.org/wiki/CJK_Unified_Ideographs_(Unicode_block)
|
| 268 |
+
#
|
| 269 |
+
# Note that the CJK Unicode block is NOT all Japanese and Korean characters,
|
| 270 |
+
# despite its name. The modern Korean Hangul alphabet is a different block,
|
| 271 |
+
# as is Japanese Hiragana and Katakana. Those alphabets are used to write
|
| 272 |
+
# space-separated words, so they are not treated specially and handled
|
| 273 |
+
# like the all of the other languages.
|
| 274 |
+
if ((cp >= 0x4E00 and cp <= 0x9FFF) or #
|
| 275 |
+
(cp >= 0x3400 and cp <= 0x4DBF) or #
|
| 276 |
+
(cp >= 0x20000 and cp <= 0x2A6DF) or #
|
| 277 |
+
(cp >= 0x2A700 and cp <= 0x2B73F) or #
|
| 278 |
+
(cp >= 0x2B740 and cp <= 0x2B81F) or #
|
| 279 |
+
(cp >= 0x2B820 and cp <= 0x2CEAF) or
|
| 280 |
+
(cp >= 0xF900 and cp <= 0xFAFF) or #
|
| 281 |
+
(cp >= 0x2F800 and cp <= 0x2FA1F)): #
|
| 282 |
+
return True
|
| 283 |
+
|
| 284 |
+
return False
|
| 285 |
+
|
| 286 |
+
def _clean_text(self, text):
|
| 287 |
+
"""Performs invalid character removal and whitespace cleanup on text."""
|
| 288 |
+
output = []
|
| 289 |
+
for char in text:
|
| 290 |
+
cp = ord(char)
|
| 291 |
+
if cp == 0 or cp == 0xfffd or _is_control(char):
|
| 292 |
+
continue
|
| 293 |
+
if _is_whitespace(char):
|
| 294 |
+
output.append(" ")
|
| 295 |
+
else:
|
| 296 |
+
output.append(char)
|
| 297 |
+
return "".join(output)
|
| 298 |
+
|
| 299 |
+
|
| 300 |
+
class WordpieceTokenizer(object):
|
| 301 |
+
"""Runs WordPiece tokenziation."""
|
| 302 |
+
|
| 303 |
+
def __init__(self, vocab, unk_token="[UNK]", max_input_chars_per_word=200):
|
| 304 |
+
self.vocab = vocab
|
| 305 |
+
self.unk_token = unk_token
|
| 306 |
+
self.max_input_chars_per_word = max_input_chars_per_word
|
| 307 |
+
|
| 308 |
+
def tokenize(self, text):
|
| 309 |
+
"""Tokenizes a piece of text into its word pieces.
|
| 310 |
+
|
| 311 |
+
This uses a greedy longest-match-first algorithm to perform tokenization
|
| 312 |
+
using the given vocabulary.
|
| 313 |
+
|
| 314 |
+
For example:
|
| 315 |
+
input = "unaffable"
|
| 316 |
+
output = ["un", "##aff", "##able"]
|
| 317 |
+
|
| 318 |
+
Args:
|
| 319 |
+
text: A single token or whitespace separated tokens. This should have
|
| 320 |
+
already been passed through `BasicTokenizer.
|
| 321 |
+
|
| 322 |
+
Returns:
|
| 323 |
+
A list of wordpiece tokens.
|
| 324 |
+
"""
|
| 325 |
+
|
| 326 |
+
text = convert_to_unicode(text)
|
| 327 |
+
|
| 328 |
+
output_tokens = []
|
| 329 |
+
for token in whitespace_tokenize(text):
|
| 330 |
+
chars = list(token)
|
| 331 |
+
if len(chars) > self.max_input_chars_per_word:
|
| 332 |
+
output_tokens.append(self.unk_token)
|
| 333 |
+
continue
|
| 334 |
+
|
| 335 |
+
is_bad = False
|
| 336 |
+
start = 0
|
| 337 |
+
sub_tokens = []
|
| 338 |
+
while start < len(chars):
|
| 339 |
+
end = len(chars)
|
| 340 |
+
cur_substr = None
|
| 341 |
+
while start < end:
|
| 342 |
+
substr = "".join(chars[start:end])
|
| 343 |
+
if start > 0:
|
| 344 |
+
substr = "##" + substr
|
| 345 |
+
if substr in self.vocab:
|
| 346 |
+
cur_substr = substr
|
| 347 |
+
break
|
| 348 |
+
end -= 1
|
| 349 |
+
if cur_substr is None:
|
| 350 |
+
is_bad = True
|
| 351 |
+
break
|
| 352 |
+
sub_tokens.append(cur_substr)
|
| 353 |
+
start = end
|
| 354 |
+
|
| 355 |
+
if is_bad:
|
| 356 |
+
output_tokens.append(self.unk_token)
|
| 357 |
+
else:
|
| 358 |
+
output_tokens.extend(sub_tokens)
|
| 359 |
+
return output_tokens
|
| 360 |
+
|
| 361 |
+
|
| 362 |
+
def _is_whitespace(char):
|
| 363 |
+
"""Checks whether `chars` is a whitespace character."""
|
| 364 |
+
# \t, \n, and \r are technically contorl characters but we treat them
|
| 365 |
+
# as whitespace since they are generally considered as such.
|
| 366 |
+
if char == " " or char == "\t" or char == "\n" or char == "\r":
|
| 367 |
+
return True
|
| 368 |
+
cat = unicodedata.category(char)
|
| 369 |
+
if cat == "Zs":
|
| 370 |
+
return True
|
| 371 |
+
return False
|
| 372 |
+
|
| 373 |
+
|
| 374 |
+
def _is_control(char):
|
| 375 |
+
"""Checks whether `chars` is a control character."""
|
| 376 |
+
# These are technically control characters but we count them as whitespace
|
| 377 |
+
# characters.
|
| 378 |
+
if char == "\t" or char == "\n" or char == "\r":
|
| 379 |
+
return False
|
| 380 |
+
cat = unicodedata.category(char)
|
| 381 |
+
if cat in ("Cc", "Cf"):
|
| 382 |
+
return True
|
| 383 |
+
return False
|
| 384 |
+
|
| 385 |
+
|
| 386 |
+
def _is_punctuation(char):
|
| 387 |
+
"""Checks whether `chars` is a punctuation character."""
|
| 388 |
+
cp = ord(char)
|
| 389 |
+
# We treat all non-letter/number ASCII as punctuation.
|
| 390 |
+
# Characters such as "^", "$", and "`" are not in the Unicode
|
| 391 |
+
# Punctuation class but we treat them as punctuation anyways, for
|
| 392 |
+
# consistency.
|
| 393 |
+
if ((cp >= 33 and cp <= 47) or (cp >= 58 and cp <= 64) or
|
| 394 |
+
(cp >= 91 and cp <= 96) or (cp >= 123 and cp <= 126)):
|
| 395 |
+
return True
|
| 396 |
+
cat = unicodedata.category(char)
|
| 397 |
+
if cat.startswith("P"):
|
| 398 |
+
return True
|
| 399 |
+
return False
|
bert-master/bert-master/tokenization_test.py
ADDED
|
@@ -0,0 +1,137 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# coding=utf-8
|
| 2 |
+
# Copyright 2018 The Google AI Language Team Authors.
|
| 3 |
+
#
|
| 4 |
+
# Licensed under the Apache License, Version 2.0 (the "License");
|
| 5 |
+
# you may not use this file except in compliance with the License.
|
| 6 |
+
# You may obtain a copy of the License at
|
| 7 |
+
#
|
| 8 |
+
# http://www.apache.org/licenses/LICENSE-2.0
|
| 9 |
+
#
|
| 10 |
+
# Unless required by applicable law or agreed to in writing, software
|
| 11 |
+
# distributed under the License is distributed on an "AS IS" BASIS,
|
| 12 |
+
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
| 13 |
+
# See the License for the specific language governing permissions and
|
| 14 |
+
# limitations under the License.
|
| 15 |
+
from __future__ import absolute_import
|
| 16 |
+
from __future__ import division
|
| 17 |
+
from __future__ import print_function
|
| 18 |
+
|
| 19 |
+
import os
|
| 20 |
+
import tempfile
|
| 21 |
+
import tokenization
|
| 22 |
+
import six
|
| 23 |
+
import tensorflow as tf
|
| 24 |
+
|
| 25 |
+
|
| 26 |
+
class TokenizationTest(tf.test.TestCase):
|
| 27 |
+
|
| 28 |
+
def test_full_tokenizer(self):
|
| 29 |
+
vocab_tokens = [
|
| 30 |
+
"[UNK]", "[CLS]", "[SEP]", "want", "##want", "##ed", "wa", "un", "runn",
|
| 31 |
+
"##ing", ","
|
| 32 |
+
]
|
| 33 |
+
with tempfile.NamedTemporaryFile(delete=False) as vocab_writer:
|
| 34 |
+
if six.PY2:
|
| 35 |
+
vocab_writer.write("".join([x + "\n" for x in vocab_tokens]))
|
| 36 |
+
else:
|
| 37 |
+
vocab_writer.write("".join(
|
| 38 |
+
[x + "\n" for x in vocab_tokens]).encode("utf-8"))
|
| 39 |
+
|
| 40 |
+
vocab_file = vocab_writer.name
|
| 41 |
+
|
| 42 |
+
tokenizer = tokenization.FullTokenizer(vocab_file)
|
| 43 |
+
os.unlink(vocab_file)
|
| 44 |
+
|
| 45 |
+
tokens = tokenizer.tokenize(u"UNwant\u00E9d,running")
|
| 46 |
+
self.assertAllEqual(tokens, ["un", "##want", "##ed", ",", "runn", "##ing"])
|
| 47 |
+
|
| 48 |
+
self.assertAllEqual(
|
| 49 |
+
tokenizer.convert_tokens_to_ids(tokens), [7, 4, 5, 10, 8, 9])
|
| 50 |
+
|
| 51 |
+
def test_chinese(self):
|
| 52 |
+
tokenizer = tokenization.BasicTokenizer()
|
| 53 |
+
|
| 54 |
+
self.assertAllEqual(
|
| 55 |
+
tokenizer.tokenize(u"ah\u535A\u63A8zz"),
|
| 56 |
+
[u"ah", u"\u535A", u"\u63A8", u"zz"])
|
| 57 |
+
|
| 58 |
+
def test_basic_tokenizer_lower(self):
|
| 59 |
+
tokenizer = tokenization.BasicTokenizer(do_lower_case=True)
|
| 60 |
+
|
| 61 |
+
self.assertAllEqual(
|
| 62 |
+
tokenizer.tokenize(u" \tHeLLo!how \n Are yoU? "),
|
| 63 |
+
["hello", "!", "how", "are", "you", "?"])
|
| 64 |
+
self.assertAllEqual(tokenizer.tokenize(u"H\u00E9llo"), ["hello"])
|
| 65 |
+
|
| 66 |
+
def test_basic_tokenizer_no_lower(self):
|
| 67 |
+
tokenizer = tokenization.BasicTokenizer(do_lower_case=False)
|
| 68 |
+
|
| 69 |
+
self.assertAllEqual(
|
| 70 |
+
tokenizer.tokenize(u" \tHeLLo!how \n Are yoU? "),
|
| 71 |
+
["HeLLo", "!", "how", "Are", "yoU", "?"])
|
| 72 |
+
|
| 73 |
+
def test_wordpiece_tokenizer(self):
|
| 74 |
+
vocab_tokens = [
|
| 75 |
+
"[UNK]", "[CLS]", "[SEP]", "want", "##want", "##ed", "wa", "un", "runn",
|
| 76 |
+
"##ing"
|
| 77 |
+
]
|
| 78 |
+
|
| 79 |
+
vocab = {}
|
| 80 |
+
for (i, token) in enumerate(vocab_tokens):
|
| 81 |
+
vocab[token] = i
|
| 82 |
+
tokenizer = tokenization.WordpieceTokenizer(vocab=vocab)
|
| 83 |
+
|
| 84 |
+
self.assertAllEqual(tokenizer.tokenize(""), [])
|
| 85 |
+
|
| 86 |
+
self.assertAllEqual(
|
| 87 |
+
tokenizer.tokenize("unwanted running"),
|
| 88 |
+
["un", "##want", "##ed", "runn", "##ing"])
|
| 89 |
+
|
| 90 |
+
self.assertAllEqual(
|
| 91 |
+
tokenizer.tokenize("unwantedX running"), ["[UNK]", "runn", "##ing"])
|
| 92 |
+
|
| 93 |
+
def test_convert_tokens_to_ids(self):
|
| 94 |
+
vocab_tokens = [
|
| 95 |
+
"[UNK]", "[CLS]", "[SEP]", "want", "##want", "##ed", "wa", "un", "runn",
|
| 96 |
+
"##ing"
|
| 97 |
+
]
|
| 98 |
+
|
| 99 |
+
vocab = {}
|
| 100 |
+
for (i, token) in enumerate(vocab_tokens):
|
| 101 |
+
vocab[token] = i
|
| 102 |
+
|
| 103 |
+
self.assertAllEqual(
|
| 104 |
+
tokenization.convert_tokens_to_ids(
|
| 105 |
+
vocab, ["un", "##want", "##ed", "runn", "##ing"]), [7, 4, 5, 8, 9])
|
| 106 |
+
|
| 107 |
+
def test_is_whitespace(self):
|
| 108 |
+
self.assertTrue(tokenization._is_whitespace(u" "))
|
| 109 |
+
self.assertTrue(tokenization._is_whitespace(u"\t"))
|
| 110 |
+
self.assertTrue(tokenization._is_whitespace(u"\r"))
|
| 111 |
+
self.assertTrue(tokenization._is_whitespace(u"\n"))
|
| 112 |
+
self.assertTrue(tokenization._is_whitespace(u"\u00A0"))
|
| 113 |
+
|
| 114 |
+
self.assertFalse(tokenization._is_whitespace(u"A"))
|
| 115 |
+
self.assertFalse(tokenization._is_whitespace(u"-"))
|
| 116 |
+
|
| 117 |
+
def test_is_control(self):
|
| 118 |
+
self.assertTrue(tokenization._is_control(u"\u0005"))
|
| 119 |
+
|
| 120 |
+
self.assertFalse(tokenization._is_control(u"A"))
|
| 121 |
+
self.assertFalse(tokenization._is_control(u" "))
|
| 122 |
+
self.assertFalse(tokenization._is_control(u"\t"))
|
| 123 |
+
self.assertFalse(tokenization._is_control(u"\r"))
|
| 124 |
+
self.assertFalse(tokenization._is_control(u"\U0001F4A9"))
|
| 125 |
+
|
| 126 |
+
def test_is_punctuation(self):
|
| 127 |
+
self.assertTrue(tokenization._is_punctuation(u"-"))
|
| 128 |
+
self.assertTrue(tokenization._is_punctuation(u"$"))
|
| 129 |
+
self.assertTrue(tokenization._is_punctuation(u"`"))
|
| 130 |
+
self.assertTrue(tokenization._is_punctuation(u"."))
|
| 131 |
+
|
| 132 |
+
self.assertFalse(tokenization._is_punctuation(u"A"))
|
| 133 |
+
self.assertFalse(tokenization._is_punctuation(u" "))
|
| 134 |
+
|
| 135 |
+
|
| 136 |
+
if __name__ == "__main__":
|
| 137 |
+
tf.test.main()
|
dark-bert-master/dark-bert-master/LICENSE
ADDED
|
@@ -0,0 +1,201 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
Apache License
|
| 2 |
+
Version 2.0, January 2004
|
| 3 |
+
http://www.apache.org/licenses/
|
| 4 |
+
|
| 5 |
+
TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
|
| 6 |
+
|
| 7 |
+
1. Definitions.
|
| 8 |
+
|
| 9 |
+
"License" shall mean the terms and conditions for use, reproduction,
|
| 10 |
+
and distribution as defined by Sections 1 through 9 of this document.
|
| 11 |
+
|
| 12 |
+
"Licensor" shall mean the copyright owner or entity authorized by
|
| 13 |
+
the copyright owner that is granting the License.
|
| 14 |
+
|
| 15 |
+
"Legal Entity" shall mean the union of the acting entity and all
|
| 16 |
+
other entities that control, are controlled by, or are under common
|
| 17 |
+
control with that entity. For the purposes of this definition,
|
| 18 |
+
"control" means (i) the power, direct or indirect, to cause the
|
| 19 |
+
direction or management of such entity, whether by contract or
|
| 20 |
+
otherwise, or (ii) ownership of fifty percent (50%) or more of the
|
| 21 |
+
outstanding shares, or (iii) beneficial ownership of such entity.
|
| 22 |
+
|
| 23 |
+
"You" (or "Your") shall mean an individual or Legal Entity
|
| 24 |
+
exercising permissions granted by this License.
|
| 25 |
+
|
| 26 |
+
"Source" form shall mean the preferred form for making modifications,
|
| 27 |
+
including but not limited to software source code, documentation
|
| 28 |
+
source, and configuration files.
|
| 29 |
+
|
| 30 |
+
"Object" form shall mean any form resulting from mechanical
|
| 31 |
+
transformation or translation of a Source form, including but
|
| 32 |
+
not limited to compiled object code, generated documentation,
|
| 33 |
+
and conversions to other media types.
|
| 34 |
+
|
| 35 |
+
"Work" shall mean the work of authorship, whether in Source or
|
| 36 |
+
Object form, made available under the License, as indicated by a
|
| 37 |
+
copyright notice that is included in or attached to the work
|
| 38 |
+
(an example is provided in the Appendix below).
|
| 39 |
+
|
| 40 |
+
"Derivative Works" shall mean any work, whether in Source or Object
|
| 41 |
+
form, that is based on (or derived from) the Work and for which the
|
| 42 |
+
editorial revisions, annotations, elaborations, or other modifications
|
| 43 |
+
represent, as a whole, an original work of authorship. For the purposes
|
| 44 |
+
of this License, Derivative Works shall not include works that remain
|
| 45 |
+
separable from, or merely link (or bind by name) to the interfaces of,
|
| 46 |
+
the Work and Derivative Works thereof.
|
| 47 |
+
|
| 48 |
+
"Contribution" shall mean any work of authorship, including
|
| 49 |
+
the original version of the Work and any modifications or additions
|
| 50 |
+
to that Work or Derivative Works thereof, that is intentionally
|
| 51 |
+
submitted to Licensor for inclusion in the Work by the copyright owner
|
| 52 |
+
or by an individual or Legal Entity authorized to submit on behalf of
|
| 53 |
+
the copyright owner. For the purposes of this definition, "submitted"
|
| 54 |
+
means any form of electronic, verbal, or written communication sent
|
| 55 |
+
to the Licensor or its representatives, including but not limited to
|
| 56 |
+
communication on electronic mailing lists, source code control systems,
|
| 57 |
+
and issue tracking systems that are managed by, or on behalf of, the
|
| 58 |
+
Licensor for the purpose of discussing and improving the Work, but
|
| 59 |
+
excluding communication that is conspicuously marked or otherwise
|
| 60 |
+
designated in writing by the copyright owner as "Not a Contribution."
|
| 61 |
+
|
| 62 |
+
"Contributor" shall mean Licensor and any individual or Legal Entity
|
| 63 |
+
on behalf of whom a Contribution has been received by Licensor and
|
| 64 |
+
subsequently incorporated within the Work.
|
| 65 |
+
|
| 66 |
+
2. Grant of Copyright License. Subject to the terms and conditions of
|
| 67 |
+
this License, each Contributor hereby grants to You a perpetual,
|
| 68 |
+
worldwide, non-exclusive, no-charge, royalty-free, irrevocable
|
| 69 |
+
copyright license to reproduce, prepare Derivative Works of,
|
| 70 |
+
publicly display, publicly perform, sublicense, and distribute the
|
| 71 |
+
Work and such Derivative Works in Source or Object form.
|
| 72 |
+
|
| 73 |
+
3. Grant of Patent License. Subject to the terms and conditions of
|
| 74 |
+
this License, each Contributor hereby grants to You a perpetual,
|
| 75 |
+
worldwide, non-exclusive, no-charge, royalty-free, irrevocable
|
| 76 |
+
(except as stated in this section) patent license to make, have made,
|
| 77 |
+
use, offer to sell, sell, import, and otherwise transfer the Work,
|
| 78 |
+
where such license applies only to those patent claims licensable
|
| 79 |
+
by such Contributor that are necessarily infringed by their
|
| 80 |
+
Contribution(s) alone or by combination of their Contribution(s)
|
| 81 |
+
with the Work to which such Contribution(s) was submitted. If You
|
| 82 |
+
institute patent litigation against any entity (including a
|
| 83 |
+
cross-claim or counterclaim in a lawsuit) alleging that the Work
|
| 84 |
+
or a Contribution incorporated within the Work constitutes direct
|
| 85 |
+
or contributory patent infringement, then any patent licenses
|
| 86 |
+
granted to You under this License for that Work shall terminate
|
| 87 |
+
as of the date such litigation is filed.
|
| 88 |
+
|
| 89 |
+
4. Redistribution. You may reproduce and distribute copies of the
|
| 90 |
+
Work or Derivative Works thereof in any medium, with or without
|
| 91 |
+
modifications, and in Source or Object form, provided that You
|
| 92 |
+
meet the following conditions:
|
| 93 |
+
|
| 94 |
+
(a) You must give any other recipients of the Work or
|
| 95 |
+
Derivative Works a copy of this License; and
|
| 96 |
+
|
| 97 |
+
(b) You must cause any modified files to carry prominent notices
|
| 98 |
+
stating that You changed the files; and
|
| 99 |
+
|
| 100 |
+
(c) You must retain, in the Source form of any Derivative Works
|
| 101 |
+
that You distribute, all copyright, patent, trademark, and
|
| 102 |
+
attribution notices from the Source form of the Work,
|
| 103 |
+
excluding those notices that do not pertain to any part of
|
| 104 |
+
the Derivative Works; and
|
| 105 |
+
|
| 106 |
+
(d) If the Work includes a "NOTICE" text file as part of its
|
| 107 |
+
distribution, then any Derivative Works that You distribute must
|
| 108 |
+
include a readable copy of the attribution notices contained
|
| 109 |
+
within such NOTICE file, excluding those notices that do not
|
| 110 |
+
pertain to any part of the Derivative Works, in at least one
|
| 111 |
+
of the following places: within a NOTICE text file distributed
|
| 112 |
+
as part of the Derivative Works; within the Source form or
|
| 113 |
+
documentation, if provided along with the Derivative Works; or,
|
| 114 |
+
within a display generated by the Derivative Works, if and
|
| 115 |
+
wherever such third-party notices normally appear. The contents
|
| 116 |
+
of the NOTICE file are for informational purposes only and
|
| 117 |
+
do not modify the License. You may add Your own attribution
|
| 118 |
+
notices within Derivative Works that You distribute, alongside
|
| 119 |
+
or as an addendum to the NOTICE text from the Work, provided
|
| 120 |
+
that such additional attribution notices cannot be construed
|
| 121 |
+
as modifying the License.
|
| 122 |
+
|
| 123 |
+
You may add Your own copyright statement to Your modifications and
|
| 124 |
+
may provide additional or different license terms and conditions
|
| 125 |
+
for use, reproduction, or distribution of Your modifications, or
|
| 126 |
+
for any such Derivative Works as a whole, provided Your use,
|
| 127 |
+
reproduction, and distribution of the Work otherwise complies with
|
| 128 |
+
the conditions stated in this License.
|
| 129 |
+
|
| 130 |
+
5. Submission of Contributions. Unless You explicitly state otherwise,
|
| 131 |
+
any Contribution intentionally submitted for inclusion in the Work
|
| 132 |
+
by You to the Licensor shall be under the terms and conditions of
|
| 133 |
+
this License, without any additional terms or conditions.
|
| 134 |
+
Notwithstanding the above, nothing herein shall supersede or modify
|
| 135 |
+
the terms of any separate license agreement you may have executed
|
| 136 |
+
with Licensor regarding such Contributions.
|
| 137 |
+
|
| 138 |
+
6. Trademarks. This License does not grant permission to use the trade
|
| 139 |
+
names, trademarks, service marks, or product names of the Licensor,
|
| 140 |
+
except as required for reasonable and customary use in describing the
|
| 141 |
+
origin of the Work and reproducing the content of the NOTICE file.
|
| 142 |
+
|
| 143 |
+
7. Disclaimer of Warranty. Unless required by applicable law or
|
| 144 |
+
agreed to in writing, Licensor provides the Work (and each
|
| 145 |
+
Contributor provides its Contributions) on an "AS IS" BASIS,
|
| 146 |
+
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
|
| 147 |
+
implied, including, without limitation, any warranties or conditions
|
| 148 |
+
of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
|
| 149 |
+
PARTICULAR PURPOSE. You are solely responsible for determining the
|
| 150 |
+
appropriateness of using or redistributing the Work and assume any
|
| 151 |
+
risks associated with Your exercise of permissions under this License.
|
| 152 |
+
|
| 153 |
+
8. Limitation of Liability. In no event and under no legal theory,
|
| 154 |
+
whether in tort (including negligence), contract, or otherwise,
|
| 155 |
+
unless required by applicable law (such as deliberate and grossly
|
| 156 |
+
negligent acts) or agreed to in writing, shall any Contributor be
|
| 157 |
+
liable to You for damages, including any direct, indirect, special,
|
| 158 |
+
incidental, or consequential damages of any character arising as a
|
| 159 |
+
result of this License or out of the use or inability to use the
|
| 160 |
+
Work (including but not limited to damages for loss of goodwill,
|
| 161 |
+
work stoppage, computer failure or malfunction, or any and all
|
| 162 |
+
other commercial damages or losses), even if such Contributor
|
| 163 |
+
has been advised of the possibility of such damages.
|
| 164 |
+
|
| 165 |
+
9. Accepting Warranty or Additional Liability. While redistributing
|
| 166 |
+
the Work or Derivative Works thereof, You may choose to offer,
|
| 167 |
+
and charge a fee for, acceptance of support, warranty, indemnity,
|
| 168 |
+
or other liability obligations and/or rights consistent with this
|
| 169 |
+
License. However, in accepting such obligations, You may act only
|
| 170 |
+
on Your own behalf and on Your sole responsibility, not on behalf
|
| 171 |
+
of any other Contributor, and only if You agree to indemnify,
|
| 172 |
+
defend, and hold each Contributor harmless for any liability
|
| 173 |
+
incurred by, or claims asserted against, such Contributor by reason
|
| 174 |
+
of your accepting any such warranty or additional liability.
|
| 175 |
+
|
| 176 |
+
END OF TERMS AND CONDITIONS
|
| 177 |
+
|
| 178 |
+
APPENDIX: How to apply the Apache License to your work.
|
| 179 |
+
|
| 180 |
+
To apply the Apache License to your work, attach the following
|
| 181 |
+
boilerplate notice, with the fields enclosed by brackets "[]"
|
| 182 |
+
replaced with your own identifying information. (Don't include
|
| 183 |
+
the brackets!) The text should be enclosed in the appropriate
|
| 184 |
+
comment syntax for the file format. We also recommend that a
|
| 185 |
+
file or class name and description of purpose be included on the
|
| 186 |
+
same "printed page" as the copyright notice for easier
|
| 187 |
+
identification within third-party archives.
|
| 188 |
+
|
| 189 |
+
Copyright [yyyy] [name of copyright owner]
|
| 190 |
+
|
| 191 |
+
Licensed under the Apache License, Version 2.0 (the "License");
|
| 192 |
+
you may not use this file except in compliance with the License.
|
| 193 |
+
You may obtain a copy of the License at
|
| 194 |
+
|
| 195 |
+
http://www.apache.org/licenses/LICENSE-2.0
|
| 196 |
+
|
| 197 |
+
Unless required by applicable law or agreed to in writing, software
|
| 198 |
+
distributed under the License is distributed on an "AS IS" BASIS,
|
| 199 |
+
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
| 200 |
+
See the License for the specific language governing permissions and
|
| 201 |
+
limitations under the License.
|
dark-bert-master/dark-bert-master/README.md
ADDED
|
@@ -0,0 +1,14 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# dark-bert
|
| 2 |
+
🧠 Using large language models to classify dark net documents in a zero-shot learning enviornments.
|
| 3 |
+
|
| 4 |
+
Dark bert eneables you to cluster any corpus of markup documents in an entirely unsupervised way.
|
| 5 |
+
|
| 6 |
+
```
|
| 7 |
+
usage: darkbert.py [-h] -m {bert,albert,roberta} -i INPUT -o OUTPUT
|
| 8 |
+
|
| 9 |
+
optional arguments:
|
| 10 |
+
-h, --help show this help message and exit
|
| 11 |
+
-m {bert,albert,roberta}, --model {bert,albert,roberta}
|
| 12 |
+
-i INPUT, --input INPUT
|
| 13 |
+
-o OUTPUT, --output OUTPUT
|
| 14 |
+
```
|
dark-bert-master/dark-bert-master/darkbert.py
ADDED
|
@@ -0,0 +1,151 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# Copyright 2022 Christopher K. Schmitt
|
| 2 |
+
#
|
| 3 |
+
# Licensed under the Apache License, Version 2.0 (the "License");
|
| 4 |
+
# you may not use this file except in compliance with the License.
|
| 5 |
+
# You may obtain a copy of the License at
|
| 6 |
+
#
|
| 7 |
+
# http://www.apache.org/licenses/LICENSE-2.0
|
| 8 |
+
#
|
| 9 |
+
# Unless required by applicable law or agreed to in writing, software
|
| 10 |
+
# distributed under the License is distributed on an "AS IS" BASIS,
|
| 11 |
+
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
| 12 |
+
# See the License for the specific language governing permissions and
|
| 13 |
+
# limitations under the License.
|
| 14 |
+
|
| 15 |
+
from sentence_transformers import SentenceTransformer
|
| 16 |
+
from sklearn.manifold import TSNE
|
| 17 |
+
from sklearn.cluster import DBSCAN
|
| 18 |
+
from sklearn.metrics import silhouette_score, calinski_harabasz_score
|
| 19 |
+
from pathlib import Path
|
| 20 |
+
from bs4 import BeautifulSoup
|
| 21 |
+
from argparse import ArgumentParser
|
| 22 |
+
|
| 23 |
+
import matplotlib.pyplot as plt
|
| 24 |
+
import numpy as np
|
| 25 |
+
import nltk as nltk
|
| 26 |
+
|
| 27 |
+
# The list of huggingface transformers with tensorflow
|
| 28 |
+
# support and compatible tokenizers.
|
| 29 |
+
available_models = {
|
| 30 |
+
"bert": "sentence-transformers/multi-qa-distilbert-cos-v1",
|
| 31 |
+
"albert": "sentence-transformers/paraphrase-albert-small-v2",
|
| 32 |
+
"roberta": "sentence-transformers/all-distilroberta-v1",
|
| 33 |
+
}
|
| 34 |
+
|
| 35 |
+
display_titles = {
|
| 36 |
+
"bert": "BERT",
|
| 37 |
+
"albert": "ALBERT",
|
| 38 |
+
"roberta": "RoBERTa",
|
| 39 |
+
}
|
| 40 |
+
|
| 41 |
+
# Define the CLI interface for modeling our data with
|
| 42 |
+
# different transformer models. We want to control the
|
| 43 |
+
# type of the tokenizer and the transformer we use, as well
|
| 44 |
+
# as the input and output directories
|
| 45 |
+
parser = ArgumentParser()
|
| 46 |
+
parser.add_argument("-m", "--model", choices=available_models.keys(), required=True)
|
| 47 |
+
parser.add_argument("-i", "--input", required=True)
|
| 48 |
+
parser.add_argument("-o", "--output", required=True)
|
| 49 |
+
|
| 50 |
+
args = parser.parse_args()
|
| 51 |
+
input_dir = args.input
|
| 52 |
+
output_dir = args.output
|
| 53 |
+
model_name = available_models[args.model]
|
| 54 |
+
display_name = display_titles[args.model]
|
| 55 |
+
|
| 56 |
+
# To remove random glyphs and other noise, we
|
| 57 |
+
# only extract words in the nltk corpus
|
| 58 |
+
nltk.download("words")
|
| 59 |
+
words = set(nltk.corpus.words.words())
|
| 60 |
+
|
| 61 |
+
def extract_words(document):
|
| 62 |
+
cleaned = ""
|
| 63 |
+
|
| 64 |
+
for word in nltk.wordpunct_tokenize(document):
|
| 65 |
+
if word.lower() in words:
|
| 66 |
+
cleaned += word.lower() + " "
|
| 67 |
+
|
| 68 |
+
return cleaned
|
| 69 |
+
|
| 70 |
+
# Iterate over all of the files in the provided data
|
| 71 |
+
# directory. Parse each file with beautiful soup to parse
|
| 72 |
+
# the relevant text out of the markup.
|
| 73 |
+
data = Path(input_dir).iterdir()
|
| 74 |
+
data = map(lambda doc: doc.read_bytes(), data)
|
| 75 |
+
data = map(lambda doc: BeautifulSoup(doc, "html.parser"), data)
|
| 76 |
+
data = map(lambda doc: doc.get_text(), data)
|
| 77 |
+
data = filter(lambda doc: len(doc) > 0, data)
|
| 78 |
+
data = map(extract_words, data)
|
| 79 |
+
data = filter(lambda doc: len(doc) > 10, data)
|
| 80 |
+
data = list(data)
|
| 81 |
+
|
| 82 |
+
# Initilize transformer models and predict all of the
|
| 83 |
+
# document embeddings as computed by bert and friends
|
| 84 |
+
model = SentenceTransformer(model_name)
|
| 85 |
+
embeddings = model.encode(data, show_progress_bar=True)
|
| 86 |
+
|
| 87 |
+
# Fit TSNE model for embedding space. Sqush down to 2
|
| 88 |
+
# dimentions for visualization purposes.
|
| 89 |
+
tsne = TSNE(n_components=2, random_state=2, init="pca", learning_rate="auto", perplexity=40)
|
| 90 |
+
tsne = tsne.fit_transform(embeddings)
|
| 91 |
+
|
| 92 |
+
# Hyperparameter optimizations
|
| 93 |
+
silhouettes = []
|
| 94 |
+
outliers = []
|
| 95 |
+
ch = []
|
| 96 |
+
|
| 97 |
+
for eps in np.arange(0.001, 1, 0.001):
|
| 98 |
+
dbscan = DBSCAN(eps, metric="cosine", n_jobs=-1)
|
| 99 |
+
dbscan = dbscan.fit_predict(embeddings)
|
| 100 |
+
|
| 101 |
+
if len(np.unique(dbscan)) > 1:
|
| 102 |
+
silhouettes.append(silhouette_score(embeddings, dbscan, metric="cosine"))
|
| 103 |
+
ch.append(calinski_harabasz_score(embeddings, dbscan))
|
| 104 |
+
else:
|
| 105 |
+
silhouettes.append(0)
|
| 106 |
+
ch.append(0)
|
| 107 |
+
|
| 108 |
+
outliers.append(len(dbscan[dbscan == -1]))
|
| 109 |
+
|
| 110 |
+
for p in range(15, 51):
|
| 111 |
+
best = np.argmax(silhouettes)
|
| 112 |
+
|
| 113 |
+
dbscan = DBSCAN(0.001 + 0.001 * best, metric="cosine", n_jobs=-1)
|
| 114 |
+
dbscan = dbscan.fit_predict(embeddings)
|
| 115 |
+
|
| 116 |
+
tsne = TSNE(n_components=2, perplexity=p, learning_rate="auto", init="pca", metric="cosine")
|
| 117 |
+
tsne = tsne.fit_transform(embeddings)
|
| 118 |
+
|
| 119 |
+
plt.figure()
|
| 120 |
+
plt.scatter(tsne[dbscan != -1][:, 0], tsne[dbscan != -1][:, 1], s=0.5, c=dbscan[dbscan != -1], cmap="hsv")
|
| 121 |
+
plt.scatter(tsne[dbscan == -1][:, 0], tsne[dbscan == -1][:, 1], s=0.5, c="#abb8c3")
|
| 122 |
+
plt.title(f"{display_name} Embeddings Visualized with T-SNE (p = {p})")
|
| 123 |
+
plt.savefig(f"{output_dir}/tnse_{p:02}.png", format="png", dpi=600)
|
| 124 |
+
plt.close()
|
| 125 |
+
|
| 126 |
+
plt.figure()
|
| 127 |
+
plt.plot(np.arange(0.001, 1, 0.001), silhouettes, lw=0.5, color="#dc322f")
|
| 128 |
+
plt.legend()
|
| 129 |
+
plt.xlabel("Epsilon")
|
| 130 |
+
plt.ylabel("silhouette score")
|
| 131 |
+
plt.title("Optimizing Epsilon by Silhouette Score")
|
| 132 |
+
plt.savefig(f"silhouettes.png", format="png", dpi=600)
|
| 133 |
+
plt.close()
|
| 134 |
+
|
| 135 |
+
plt.figure()
|
| 136 |
+
plt.plot(np.arange(0.001, 1, 0.001), outliers, lw=0.5, color="#dc322f")
|
| 137 |
+
plt.legend()
|
| 138 |
+
plt.xlabel("Epsilon")
|
| 139 |
+
plt.ylabel("outliers")
|
| 140 |
+
plt.title("Optimizing Epsilon by Number of Outliers")
|
| 141 |
+
plt.savefig(f"outliers.png", format="png", dpi=600)
|
| 142 |
+
plt.close()
|
| 143 |
+
|
| 144 |
+
plt.figure()
|
| 145 |
+
plt.plot(np.arange(0.001, 1, 0.001), ch, lw=0.5, color="#dc322f")
|
| 146 |
+
plt.legend()
|
| 147 |
+
plt.xlabel("Epsilon")
|
| 148 |
+
plt.ylabel("Calinski-Harabasz score")
|
| 149 |
+
plt.title("Optimizing Epsilon by Calinski-Harabasz Score")
|
| 150 |
+
plt.savefig(f"calinski-harabasz.png", format="png", dpi=600)
|
| 151 |
+
plt.close()
|
dark-bert-master/dark-bert-master/requirements.txt
ADDED
|
Binary file (1.54 kB). View file
|
|
|
dark-secrets-of-BERT-master/dark-secrets-of-BERT-master.gitignore
ADDED
|
@@ -0,0 +1,125 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# Initially taken from Github's Python gitignore file
|
| 2 |
+
|
| 3 |
+
# Byte-compiled / optimized / DLL files
|
| 4 |
+
__pycache__/
|
| 5 |
+
*.py[cod]
|
| 6 |
+
*$py.class
|
| 7 |
+
|
| 8 |
+
# C extensions
|
| 9 |
+
*.so
|
| 10 |
+
|
| 11 |
+
# Distribution / packaging
|
| 12 |
+
.Python
|
| 13 |
+
build/
|
| 14 |
+
develop-eggs/
|
| 15 |
+
dist/
|
| 16 |
+
downloads/
|
| 17 |
+
eggs/
|
| 18 |
+
.eggs/
|
| 19 |
+
lib/
|
| 20 |
+
lib64/
|
| 21 |
+
parts/
|
| 22 |
+
sdist/
|
| 23 |
+
var/
|
| 24 |
+
wheels/
|
| 25 |
+
*.egg-info/
|
| 26 |
+
.installed.cfg
|
| 27 |
+
*.egg
|
| 28 |
+
MANIFEST
|
| 29 |
+
|
| 30 |
+
# PyInstaller
|
| 31 |
+
# Usually these files are written by a python script from a template
|
| 32 |
+
# before PyInstaller builds the exe, so as to inject date/other infos into it.
|
| 33 |
+
*.manifest
|
| 34 |
+
*.spec
|
| 35 |
+
|
| 36 |
+
# Installer logs
|
| 37 |
+
pip-log.txt
|
| 38 |
+
pip-delete-this-directory.txt
|
| 39 |
+
|
| 40 |
+
# Unit test / coverage reports
|
| 41 |
+
htmlcov/
|
| 42 |
+
.tox/
|
| 43 |
+
.nox/
|
| 44 |
+
.coverage
|
| 45 |
+
.coverage.*
|
| 46 |
+
.cache
|
| 47 |
+
nosetests.xml
|
| 48 |
+
coverage.xml
|
| 49 |
+
*.cover
|
| 50 |
+
.hypothesis/
|
| 51 |
+
.pytest_cache/
|
| 52 |
+
|
| 53 |
+
# Translations
|
| 54 |
+
*.mo
|
| 55 |
+
*.pot
|
| 56 |
+
|
| 57 |
+
# Django stuff:
|
| 58 |
+
*.log
|
| 59 |
+
local_settings.py
|
| 60 |
+
db.sqlite3
|
| 61 |
+
|
| 62 |
+
# Flask stuff:
|
| 63 |
+
instance/
|
| 64 |
+
.webassets-cache
|
| 65 |
+
|
| 66 |
+
# Scrapy stuff:
|
| 67 |
+
.scrapy
|
| 68 |
+
|
| 69 |
+
# Sphinx documentation
|
| 70 |
+
docs/_build/
|
| 71 |
+
|
| 72 |
+
# PyBuilder
|
| 73 |
+
target/
|
| 74 |
+
|
| 75 |
+
# Jupyter Notebook
|
| 76 |
+
.ipynb_checkpoints
|
| 77 |
+
|
| 78 |
+
# IPython
|
| 79 |
+
profile_default/
|
| 80 |
+
ipython_config.py
|
| 81 |
+
|
| 82 |
+
# pyenv
|
| 83 |
+
.python-version
|
| 84 |
+
|
| 85 |
+
# celery beat schedule file
|
| 86 |
+
celerybeat-schedule
|
| 87 |
+
|
| 88 |
+
# SageMath parsed files
|
| 89 |
+
*.sage.py
|
| 90 |
+
|
| 91 |
+
# Environments
|
| 92 |
+
.env
|
| 93 |
+
.venv
|
| 94 |
+
env/
|
| 95 |
+
venv/
|
| 96 |
+
ENV/
|
| 97 |
+
env.bak/
|
| 98 |
+
venv.bak/
|
| 99 |
+
|
| 100 |
+
# Spyder project settings
|
| 101 |
+
.spyderproject
|
| 102 |
+
.spyproject
|
| 103 |
+
|
| 104 |
+
# Rope project settings
|
| 105 |
+
.ropeproject
|
| 106 |
+
|
| 107 |
+
# mkdocs documentation
|
| 108 |
+
/site
|
| 109 |
+
|
| 110 |
+
# mypy
|
| 111 |
+
.mypy_cache/
|
| 112 |
+
.dmypy.json
|
| 113 |
+
dmypy.json
|
| 114 |
+
|
| 115 |
+
# Pyre type checker
|
| 116 |
+
.pyre/
|
| 117 |
+
|
| 118 |
+
# vscode
|
| 119 |
+
.vscode
|
| 120 |
+
|
| 121 |
+
# TF code
|
| 122 |
+
tensorflow_code
|
| 123 |
+
|
| 124 |
+
# Models
|
| 125 |
+
models
|
dark-secrets-of-BERT-master/dark-secrets-of-BERT-master/.circleci/config.yml
ADDED
|
@@ -0,0 +1,29 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version: 2
|
| 2 |
+
jobs:
|
| 3 |
+
build_py3:
|
| 4 |
+
working_directory: ~/pytorch-pretrained-BERT
|
| 5 |
+
docker:
|
| 6 |
+
- image: circleci/python:3.5
|
| 7 |
+
steps:
|
| 8 |
+
- checkout
|
| 9 |
+
- run: sudo pip install --progress-bar off .
|
| 10 |
+
- run: sudo pip install pytest ftfy spacy
|
| 11 |
+
- run: sudo python -m spacy download en
|
| 12 |
+
- run: python -m pytest -sv tests/ --runslow
|
| 13 |
+
build_py2:
|
| 14 |
+
working_directory: ~/pytorch-pretrained-BERT
|
| 15 |
+
docker:
|
| 16 |
+
- image: circleci/python:2.7
|
| 17 |
+
steps:
|
| 18 |
+
- checkout
|
| 19 |
+
- run: sudo pip install --progress-bar off .
|
| 20 |
+
- run: sudo pip install pytest spacy
|
| 21 |
+
- run: sudo pip install ftfy==4.4.3
|
| 22 |
+
- run: sudo python -m spacy download en
|
| 23 |
+
- run: python -m pytest -sv tests/ --runslow
|
| 24 |
+
workflows:
|
| 25 |
+
version: 2
|
| 26 |
+
build_and_test:
|
| 27 |
+
jobs:
|
| 28 |
+
- build_py3
|
| 29 |
+
- build_py2
|
dark-secrets-of-BERT-master/dark-secrets-of-BERT-master/.github/stale.yml
ADDED
|
@@ -0,0 +1,17 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# Number of days of inactivity before an issue becomes stale
|
| 2 |
+
daysUntilStale: 60
|
| 3 |
+
# Number of days of inactivity before a stale issue is closed
|
| 4 |
+
daysUntilClose: 7
|
| 5 |
+
# Issues with these labels will never be considered stale
|
| 6 |
+
exemptLabels:
|
| 7 |
+
- pinned
|
| 8 |
+
- security
|
| 9 |
+
# Label to use when marking an issue as stale
|
| 10 |
+
staleLabel: wontfix
|
| 11 |
+
# Comment to post when marking an issue as stale. Set to `false` to disable
|
| 12 |
+
markComment: >
|
| 13 |
+
This issue has been automatically marked as stale because it has not had
|
| 14 |
+
recent activity. It will be closed if no further activity occurs. Thank you
|
| 15 |
+
for your contributions.
|
| 16 |
+
# Comment to post when closing a stale issue. Set to `false` to disable
|
| 17 |
+
closeComment: false
|
dark-secrets-of-BERT-master/dark-secrets-of-BERT-master/LICENSE
ADDED
|
@@ -0,0 +1,202 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
|
| 2 |
+
Apache License
|
| 3 |
+
Version 2.0, January 2004
|
| 4 |
+
http://www.apache.org/licenses/
|
| 5 |
+
|
| 6 |
+
TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
|
| 7 |
+
|
| 8 |
+
1. Definitions.
|
| 9 |
+
|
| 10 |
+
"License" shall mean the terms and conditions for use, reproduction,
|
| 11 |
+
and distribution as defined by Sections 1 through 9 of this document.
|
| 12 |
+
|
| 13 |
+
"Licensor" shall mean the copyright owner or entity authorized by
|
| 14 |
+
the copyright owner that is granting the License.
|
| 15 |
+
|
| 16 |
+
"Legal Entity" shall mean the union of the acting entity and all
|
| 17 |
+
other entities that control, are controlled by, or are under common
|
| 18 |
+
control with that entity. For the purposes of this definition,
|
| 19 |
+
"control" means (i) the power, direct or indirect, to cause the
|
| 20 |
+
direction or management of such entity, whether by contract or
|
| 21 |
+
otherwise, or (ii) ownership of fifty percent (50%) or more of the
|
| 22 |
+
outstanding shares, or (iii) beneficial ownership of such entity.
|
| 23 |
+
|
| 24 |
+
"You" (or "Your") shall mean an individual or Legal Entity
|
| 25 |
+
exercising permissions granted by this License.
|
| 26 |
+
|
| 27 |
+
"Source" form shall mean the preferred form for making modifications,
|
| 28 |
+
including but not limited to software source code, documentation
|
| 29 |
+
source, and configuration files.
|
| 30 |
+
|
| 31 |
+
"Object" form shall mean any form resulting from mechanical
|
| 32 |
+
transformation or translation of a Source form, including but
|
| 33 |
+
not limited to compiled object code, generated documentation,
|
| 34 |
+
and conversions to other media types.
|
| 35 |
+
|
| 36 |
+
"Work" shall mean the work of authorship, whether in Source or
|
| 37 |
+
Object form, made available under the License, as indicated by a
|
| 38 |
+
copyright notice that is included in or attached to the work
|
| 39 |
+
(an example is provided in the Appendix below).
|
| 40 |
+
|
| 41 |
+
"Derivative Works" shall mean any work, whether in Source or Object
|
| 42 |
+
form, that is based on (or derived from) the Work and for which the
|
| 43 |
+
editorial revisions, annotations, elaborations, or other modifications
|
| 44 |
+
represent, as a whole, an original work of authorship. For the purposes
|
| 45 |
+
of this License, Derivative Works shall not include works that remain
|
| 46 |
+
separable from, or merely link (or bind by name) to the interfaces of,
|
| 47 |
+
the Work and Derivative Works thereof.
|
| 48 |
+
|
| 49 |
+
"Contribution" shall mean any work of authorship, including
|
| 50 |
+
the original version of the Work and any modifications or additions
|
| 51 |
+
to that Work or Derivative Works thereof, that is intentionally
|
| 52 |
+
submitted to Licensor for inclusion in the Work by the copyright owner
|
| 53 |
+
or by an individual or Legal Entity authorized to submit on behalf of
|
| 54 |
+
the copyright owner. For the purposes of this definition, "submitted"
|
| 55 |
+
means any form of electronic, verbal, or written communication sent
|
| 56 |
+
to the Licensor or its representatives, including but not limited to
|
| 57 |
+
communication on electronic mailing lists, source code control systems,
|
| 58 |
+
and issue tracking systems that are managed by, or on behalf of, the
|
| 59 |
+
Licensor for the purpose of discussing and improving the Work, but
|
| 60 |
+
excluding communication that is conspicuously marked or otherwise
|
| 61 |
+
designated in writing by the copyright owner as "Not a Contribution."
|
| 62 |
+
|
| 63 |
+
"Contributor" shall mean Licensor and any individual or Legal Entity
|
| 64 |
+
on behalf of whom a Contribution has been received by Licensor and
|
| 65 |
+
subsequently incorporated within the Work.
|
| 66 |
+
|
| 67 |
+
2. Grant of Copyright License. Subject to the terms and conditions of
|
| 68 |
+
this License, each Contributor hereby grants to You a perpetual,
|
| 69 |
+
worldwide, non-exclusive, no-charge, royalty-free, irrevocable
|
| 70 |
+
copyright license to reproduce, prepare Derivative Works of,
|
| 71 |
+
publicly display, publicly perform, sublicense, and distribute the
|
| 72 |
+
Work and such Derivative Works in Source or Object form.
|
| 73 |
+
|
| 74 |
+
3. Grant of Patent License. Subject to the terms and conditions of
|
| 75 |
+
this License, each Contributor hereby grants to You a perpetual,
|
| 76 |
+
worldwide, non-exclusive, no-charge, royalty-free, irrevocable
|
| 77 |
+
(except as stated in this section) patent license to make, have made,
|
| 78 |
+
use, offer to sell, sell, import, and otherwise transfer the Work,
|
| 79 |
+
where such license applies only to those patent claims licensable
|
| 80 |
+
by such Contributor that are necessarily infringed by their
|
| 81 |
+
Contribution(s) alone or by combination of their Contribution(s)
|
| 82 |
+
with the Work to which such Contribution(s) was submitted. If You
|
| 83 |
+
institute patent litigation against any entity (including a
|
| 84 |
+
cross-claim or counterclaim in a lawsuit) alleging that the Work
|
| 85 |
+
or a Contribution incorporated within the Work constitutes direct
|
| 86 |
+
or contributory patent infringement, then any patent licenses
|
| 87 |
+
granted to You under this License for that Work shall terminate
|
| 88 |
+
as of the date such litigation is filed.
|
| 89 |
+
|
| 90 |
+
4. Redistribution. You may reproduce and distribute copies of the
|
| 91 |
+
Work or Derivative Works thereof in any medium, with or without
|
| 92 |
+
modifications, and in Source or Object form, provided that You
|
| 93 |
+
meet the following conditions:
|
| 94 |
+
|
| 95 |
+
(a) You must give any other recipients of the Work or
|
| 96 |
+
Derivative Works a copy of this License; and
|
| 97 |
+
|
| 98 |
+
(b) You must cause any modified files to carry prominent notices
|
| 99 |
+
stating that You changed the files; and
|
| 100 |
+
|
| 101 |
+
(c) You must retain, in the Source form of any Derivative Works
|
| 102 |
+
that You distribute, all copyright, patent, trademark, and
|
| 103 |
+
attribution notices from the Source form of the Work,
|
| 104 |
+
excluding those notices that do not pertain to any part of
|
| 105 |
+
the Derivative Works; and
|
| 106 |
+
|
| 107 |
+
(d) If the Work includes a "NOTICE" text file as part of its
|
| 108 |
+
distribution, then any Derivative Works that You distribute must
|
| 109 |
+
include a readable copy of the attribution notices contained
|
| 110 |
+
within such NOTICE file, excluding those notices that do not
|
| 111 |
+
pertain to any part of the Derivative Works, in at least one
|
| 112 |
+
of the following places: within a NOTICE text file distributed
|
| 113 |
+
as part of the Derivative Works; within the Source form or
|
| 114 |
+
documentation, if provided along with the Derivative Works; or,
|
| 115 |
+
within a display generated by the Derivative Works, if and
|
| 116 |
+
wherever such third-party notices normally appear. The contents
|
| 117 |
+
of the NOTICE file are for informational purposes only and
|
| 118 |
+
do not modify the License. You may add Your own attribution
|
| 119 |
+
notices within Derivative Works that You distribute, alongside
|
| 120 |
+
or as an addendum to the NOTICE text from the Work, provided
|
| 121 |
+
that such additional attribution notices cannot be construed
|
| 122 |
+
as modifying the License.
|
| 123 |
+
|
| 124 |
+
You may add Your own copyright statement to Your modifications and
|
| 125 |
+
may provide additional or different license terms and conditions
|
| 126 |
+
for use, reproduction, or distribution of Your modifications, or
|
| 127 |
+
for any such Derivative Works as a whole, provided Your use,
|
| 128 |
+
reproduction, and distribution of the Work otherwise complies with
|
| 129 |
+
the conditions stated in this License.
|
| 130 |
+
|
| 131 |
+
5. Submission of Contributions. Unless You explicitly state otherwise,
|
| 132 |
+
any Contribution intentionally submitted for inclusion in the Work
|
| 133 |
+
by You to the Licensor shall be under the terms and conditions of
|
| 134 |
+
this License, without any additional terms or conditions.
|
| 135 |
+
Notwithstanding the above, nothing herein shall supersede or modify
|
| 136 |
+
the terms of any separate license agreement you may have executed
|
| 137 |
+
with Licensor regarding such Contributions.
|
| 138 |
+
|
| 139 |
+
6. Trademarks. This License does not grant permission to use the trade
|
| 140 |
+
names, trademarks, service marks, or product names of the Licensor,
|
| 141 |
+
except as required for reasonable and customary use in describing the
|
| 142 |
+
origin of the Work and reproducing the content of the NOTICE file.
|
| 143 |
+
|
| 144 |
+
7. Disclaimer of Warranty. Unless required by applicable law or
|
| 145 |
+
agreed to in writing, Licensor provides the Work (and each
|
| 146 |
+
Contributor provides its Contributions) on an "AS IS" BASIS,
|
| 147 |
+
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
|
| 148 |
+
implied, including, without limitation, any warranties or conditions
|
| 149 |
+
of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
|
| 150 |
+
PARTICULAR PURPOSE. You are solely responsible for determining the
|
| 151 |
+
appropriateness of using or redistributing the Work and assume any
|
| 152 |
+
risks associated with Your exercise of permissions under this License.
|
| 153 |
+
|
| 154 |
+
8. Limitation of Liability. In no event and under no legal theory,
|
| 155 |
+
whether in tort (including negligence), contract, or otherwise,
|
| 156 |
+
unless required by applicable law (such as deliberate and grossly
|
| 157 |
+
negligent acts) or agreed to in writing, shall any Contributor be
|
| 158 |
+
liable to You for damages, including any direct, indirect, special,
|
| 159 |
+
incidental, or consequential damages of any character arising as a
|
| 160 |
+
result of this License or out of the use or inability to use the
|
| 161 |
+
Work (including but not limited to damages for loss of goodwill,
|
| 162 |
+
work stoppage, computer failure or malfunction, or any and all
|
| 163 |
+
other commercial damages or losses), even if such Contributor
|
| 164 |
+
has been advised of the possibility of such damages.
|
| 165 |
+
|
| 166 |
+
9. Accepting Warranty or Additional Liability. While redistributing
|
| 167 |
+
the Work or Derivative Works thereof, You may choose to offer,
|
| 168 |
+
and charge a fee for, acceptance of support, warranty, indemnity,
|
| 169 |
+
or other liability obligations and/or rights consistent with this
|
| 170 |
+
License. However, in accepting such obligations, You may act only
|
| 171 |
+
on Your own behalf and on Your sole responsibility, not on behalf
|
| 172 |
+
of any other Contributor, and only if You agree to indemnify,
|
| 173 |
+
defend, and hold each Contributor harmless for any liability
|
| 174 |
+
incurred by, or claims asserted against, such Contributor by reason
|
| 175 |
+
of your accepting any such warranty or additional liability.
|
| 176 |
+
|
| 177 |
+
END OF TERMS AND CONDITIONS
|
| 178 |
+
|
| 179 |
+
APPENDIX: How to apply the Apache License to your work.
|
| 180 |
+
|
| 181 |
+
To apply the Apache License to your work, attach the following
|
| 182 |
+
boilerplate notice, with the fields enclosed by brackets "[]"
|
| 183 |
+
replaced with your own identifying information. (Don't include
|
| 184 |
+
the brackets!) The text should be enclosed in the appropriate
|
| 185 |
+
comment syntax for the file format. We also recommend that a
|
| 186 |
+
file or class name and description of purpose be included on the
|
| 187 |
+
same "printed page" as the copyright notice for easier
|
| 188 |
+
identification within third-party archives.
|
| 189 |
+
|
| 190 |
+
Copyright [yyyy] [name of copyright owner]
|
| 191 |
+
|
| 192 |
+
Licensed under the Apache License, Version 2.0 (the "License");
|
| 193 |
+
you may not use this file except in compliance with the License.
|
| 194 |
+
You may obtain a copy of the License at
|
| 195 |
+
|
| 196 |
+
http://www.apache.org/licenses/LICENSE-2.0
|
| 197 |
+
|
| 198 |
+
Unless required by applicable law or agreed to in writing, software
|
| 199 |
+
distributed under the License is distributed on an "AS IS" BASIS,
|
| 200 |
+
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
| 201 |
+
See the License for the specific language governing permissions and
|
| 202 |
+
limitations under the License.
|
dark-secrets-of-BERT-master/dark-secrets-of-BERT-master/MANIFEST.in
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 1 |
+
include LICENSE
|
dark-secrets-of-BERT-master/dark-secrets-of-BERT-master/README.md
ADDED
|
@@ -0,0 +1,30 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# Introduction
|
| 2 |
+
For installation and docs please refer to [release 0.6 of pytorch_pretrained_bert](https://github.com/huggingface/transformers/releases).
|
| 3 |
+
|
| 4 |
+
The current fork adds the jupyter notebook on the attention analysis of the 12 layer BERT model. For details please refer to [the paper](https://arxiv.org/abs/1908.08593).
|
| 5 |
+
|
| 6 |
+
Note that for the extraction of attention weights the source code of [](./pytorch_pretrained_bert/modeling.py) was modified (this functionality was added in later realeases of the forked repo).
|
| 7 |
+
|
| 8 |
+
# Requirements and usage
|
| 9 |
+
1. Install the requirements as
|
| 10 |
+
```
|
| 11 |
+
pip install -r requirements.txt
|
| 12 |
+
```
|
| 13 |
+
|
| 14 |
+
2. The current implementation assumes you have GLUE datasets downloaded and fine-tuned BERT model weights saved to a directory of your choice. You can download GLUE data as described [here](https://github.com/nyu-mll/GLUE-baselines/blob/master/download_glue_data.py). To fine-tune BERT, run [](./examples/run_classifier.py).
|
| 15 |
+
|
| 16 |
+
3. The code for analysis is contained in [the jupyter notebook](./visualize_attention.ipynb).
|
| 17 |
+
|
| 18 |
+
4. To repeat the results of experiments, make sure to change the `path_to_model` and `path_to_data` in the notebook.
|
| 19 |
+
|
| 20 |
+
|
| 21 |
+
# References
|
| 22 |
+
```
|
| 23 |
+
@article{kovaleva2019revealing,
|
| 24 |
+
title={Revealing the Dark Secrets of BERT},
|
| 25 |
+
author={Kovaleva, Olga and Romanov, Alexey and Rogers, Anna and Rumshisky, Anna},
|
| 26 |
+
journal={arXiv preprint arXiv:1908.08593},
|
| 27 |
+
year={2019}
|
| 28 |
+
}
|
| 29 |
+
```
|
| 30 |
+
|
dark-secrets-of-BERT-master/dark-secrets-of-BERT-master/docker/Dockerfile
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
FROM pytorch/pytorch:latest
|
| 2 |
+
|
| 3 |
+
RUN git clone https://github.com/NVIDIA/apex.git && cd apex && python setup.py install --cuda_ext --cpp_ext
|
| 4 |
+
|
| 5 |
+
RUN pip install pytorch-pretrained-bert
|
| 6 |
+
|
| 7 |
+
WORKDIR /workspace
|
dark-secrets-of-BERT-master/dark-secrets-of-BERT-master/docs/imgs/warmup_constant_schedule.png
ADDED
|
dark-secrets-of-BERT-master/dark-secrets-of-BERT-master/docs/imgs/warmup_cosine_hard_restarts_schedule.png
ADDED
|
dark-secrets-of-BERT-master/dark-secrets-of-BERT-master/docs/imgs/warmup_cosine_schedule.png
ADDED
|
dark-secrets-of-BERT-master/dark-secrets-of-BERT-master/docs/imgs/warmup_cosine_warm_restarts_schedule.png
ADDED
|
dark-secrets-of-BERT-master/dark-secrets-of-BERT-master/docs/imgs/warmup_linear_schedule.png
ADDED
|
dark-secrets-of-BERT-master/dark-secrets-of-BERT-master/examples/extract_features.py
ADDED
|
@@ -0,0 +1,297 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# coding=utf-8
|
| 2 |
+
# Copyright 2018 The Google AI Language Team Authors and The HuggingFace Inc. team.
|
| 3 |
+
#
|
| 4 |
+
# Licensed under the Apache License, Version 2.0 (the "License");
|
| 5 |
+
# you may not use this file except in compliance with the License.
|
| 6 |
+
# You may obtain a copy of the License at
|
| 7 |
+
#
|
| 8 |
+
# http://www.apache.org/licenses/LICENSE-2.0
|
| 9 |
+
#
|
| 10 |
+
# Unless required by applicable law or agreed to in writing, software
|
| 11 |
+
# distributed under the License is distributed on an "AS IS" BASIS,
|
| 12 |
+
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
| 13 |
+
# See the License for the specific language governing permissions and
|
| 14 |
+
# limitations under the License.
|
| 15 |
+
"""Extract pre-computed feature vectors from a PyTorch BERT model."""
|
| 16 |
+
|
| 17 |
+
from __future__ import absolute_import
|
| 18 |
+
from __future__ import division
|
| 19 |
+
from __future__ import print_function
|
| 20 |
+
|
| 21 |
+
import argparse
|
| 22 |
+
import collections
|
| 23 |
+
import logging
|
| 24 |
+
import json
|
| 25 |
+
import re
|
| 26 |
+
|
| 27 |
+
import torch
|
| 28 |
+
from torch.utils.data import TensorDataset, DataLoader, SequentialSampler
|
| 29 |
+
from torch.utils.data.distributed import DistributedSampler
|
| 30 |
+
|
| 31 |
+
from pytorch_pretrained_bert.tokenization import BertTokenizer
|
| 32 |
+
from pytorch_pretrained_bert.modeling import BertModel
|
| 33 |
+
|
| 34 |
+
logging.basicConfig(format = '%(asctime)s - %(levelname)s - %(name)s - %(message)s',
|
| 35 |
+
datefmt = '%m/%d/%Y %H:%M:%S',
|
| 36 |
+
level = logging.INFO)
|
| 37 |
+
logger = logging.getLogger(__name__)
|
| 38 |
+
|
| 39 |
+
|
| 40 |
+
class InputExample(object):
|
| 41 |
+
|
| 42 |
+
def __init__(self, unique_id, text_a, text_b):
|
| 43 |
+
self.unique_id = unique_id
|
| 44 |
+
self.text_a = text_a
|
| 45 |
+
self.text_b = text_b
|
| 46 |
+
|
| 47 |
+
|
| 48 |
+
class InputFeatures(object):
|
| 49 |
+
"""A single set of features of data."""
|
| 50 |
+
|
| 51 |
+
def __init__(self, unique_id, tokens, input_ids, input_mask, input_type_ids):
|
| 52 |
+
self.unique_id = unique_id
|
| 53 |
+
self.tokens = tokens
|
| 54 |
+
self.input_ids = input_ids
|
| 55 |
+
self.input_mask = input_mask
|
| 56 |
+
self.input_type_ids = input_type_ids
|
| 57 |
+
|
| 58 |
+
|
| 59 |
+
def convert_examples_to_features(examples, seq_length, tokenizer):
|
| 60 |
+
"""Loads a data file into a list of `InputFeature`s."""
|
| 61 |
+
|
| 62 |
+
features = []
|
| 63 |
+
for (ex_index, example) in enumerate(examples):
|
| 64 |
+
tokens_a = tokenizer.tokenize(example.text_a)
|
| 65 |
+
|
| 66 |
+
tokens_b = None
|
| 67 |
+
if example.text_b:
|
| 68 |
+
tokens_b = tokenizer.tokenize(example.text_b)
|
| 69 |
+
|
| 70 |
+
if tokens_b:
|
| 71 |
+
# Modifies `tokens_a` and `tokens_b` in place so that the total
|
| 72 |
+
# length is less than the specified length.
|
| 73 |
+
# Account for [CLS], [SEP], [SEP] with "- 3"
|
| 74 |
+
_truncate_seq_pair(tokens_a, tokens_b, seq_length - 3)
|
| 75 |
+
else:
|
| 76 |
+
# Account for [CLS] and [SEP] with "- 2"
|
| 77 |
+
if len(tokens_a) > seq_length - 2:
|
| 78 |
+
tokens_a = tokens_a[0:(seq_length - 2)]
|
| 79 |
+
|
| 80 |
+
# The convention in BERT is:
|
| 81 |
+
# (a) For sequence pairs:
|
| 82 |
+
# tokens: [CLS] is this jack ##son ##ville ? [SEP] no it is not . [SEP]
|
| 83 |
+
# type_ids: 0 0 0 0 0 0 0 0 1 1 1 1 1 1
|
| 84 |
+
# (b) For single sequences:
|
| 85 |
+
# tokens: [CLS] the dog is hairy . [SEP]
|
| 86 |
+
# type_ids: 0 0 0 0 0 0 0
|
| 87 |
+
#
|
| 88 |
+
# Where "type_ids" are used to indicate whether this is the first
|
| 89 |
+
# sequence or the second sequence. The embedding vectors for `type=0` and
|
| 90 |
+
# `type=1` were learned during pre-training and are added to the wordpiece
|
| 91 |
+
# embedding vector (and position vector). This is not *strictly* necessary
|
| 92 |
+
# since the [SEP] token unambigiously separates the sequences, but it makes
|
| 93 |
+
# it easier for the model to learn the concept of sequences.
|
| 94 |
+
#
|
| 95 |
+
# For classification tasks, the first vector (corresponding to [CLS]) is
|
| 96 |
+
# used as as the "sentence vector". Note that this only makes sense because
|
| 97 |
+
# the entire model is fine-tuned.
|
| 98 |
+
tokens = []
|
| 99 |
+
input_type_ids = []
|
| 100 |
+
tokens.append("[CLS]")
|
| 101 |
+
input_type_ids.append(0)
|
| 102 |
+
for token in tokens_a:
|
| 103 |
+
tokens.append(token)
|
| 104 |
+
input_type_ids.append(0)
|
| 105 |
+
tokens.append("[SEP]")
|
| 106 |
+
input_type_ids.append(0)
|
| 107 |
+
|
| 108 |
+
if tokens_b:
|
| 109 |
+
for token in tokens_b:
|
| 110 |
+
tokens.append(token)
|
| 111 |
+
input_type_ids.append(1)
|
| 112 |
+
tokens.append("[SEP]")
|
| 113 |
+
input_type_ids.append(1)
|
| 114 |
+
|
| 115 |
+
input_ids = tokenizer.convert_tokens_to_ids(tokens)
|
| 116 |
+
|
| 117 |
+
# The mask has 1 for real tokens and 0 for padding tokens. Only real
|
| 118 |
+
# tokens are attended to.
|
| 119 |
+
input_mask = [1] * len(input_ids)
|
| 120 |
+
|
| 121 |
+
# Zero-pad up to the sequence length.
|
| 122 |
+
while len(input_ids) < seq_length:
|
| 123 |
+
input_ids.append(0)
|
| 124 |
+
input_mask.append(0)
|
| 125 |
+
input_type_ids.append(0)
|
| 126 |
+
|
| 127 |
+
assert len(input_ids) == seq_length
|
| 128 |
+
assert len(input_mask) == seq_length
|
| 129 |
+
assert len(input_type_ids) == seq_length
|
| 130 |
+
|
| 131 |
+
if ex_index < 5:
|
| 132 |
+
logger.info("*** Example ***")
|
| 133 |
+
logger.info("unique_id: %s" % (example.unique_id))
|
| 134 |
+
logger.info("tokens: %s" % " ".join([str(x) for x in tokens]))
|
| 135 |
+
logger.info("input_ids: %s" % " ".join([str(x) for x in input_ids]))
|
| 136 |
+
logger.info("input_mask: %s" % " ".join([str(x) for x in input_mask]))
|
| 137 |
+
logger.info(
|
| 138 |
+
"input_type_ids: %s" % " ".join([str(x) for x in input_type_ids]))
|
| 139 |
+
|
| 140 |
+
features.append(
|
| 141 |
+
InputFeatures(
|
| 142 |
+
unique_id=example.unique_id,
|
| 143 |
+
tokens=tokens,
|
| 144 |
+
input_ids=input_ids,
|
| 145 |
+
input_mask=input_mask,
|
| 146 |
+
input_type_ids=input_type_ids))
|
| 147 |
+
return features
|
| 148 |
+
|
| 149 |
+
|
| 150 |
+
def _truncate_seq_pair(tokens_a, tokens_b, max_length):
|
| 151 |
+
"""Truncates a sequence pair in place to the maximum length."""
|
| 152 |
+
|
| 153 |
+
# This is a simple heuristic which will always truncate the longer sequence
|
| 154 |
+
# one token at a time. This makes more sense than truncating an equal percent
|
| 155 |
+
# of tokens from each, since if one sequence is very short then each token
|
| 156 |
+
# that's truncated likely contains more information than a longer sequence.
|
| 157 |
+
while True:
|
| 158 |
+
total_length = len(tokens_a) + len(tokens_b)
|
| 159 |
+
if total_length <= max_length:
|
| 160 |
+
break
|
| 161 |
+
if len(tokens_a) > len(tokens_b):
|
| 162 |
+
tokens_a.pop()
|
| 163 |
+
else:
|
| 164 |
+
tokens_b.pop()
|
| 165 |
+
|
| 166 |
+
|
| 167 |
+
def read_examples(input_file):
|
| 168 |
+
"""Read a list of `InputExample`s from an input file."""
|
| 169 |
+
examples = []
|
| 170 |
+
unique_id = 0
|
| 171 |
+
with open(input_file, "r", encoding='utf-8') as reader:
|
| 172 |
+
while True:
|
| 173 |
+
line = reader.readline()
|
| 174 |
+
if not line:
|
| 175 |
+
break
|
| 176 |
+
line = line.strip()
|
| 177 |
+
text_a = None
|
| 178 |
+
text_b = None
|
| 179 |
+
m = re.match(r"^(.*) \|\|\| (.*)$", line)
|
| 180 |
+
if m is None:
|
| 181 |
+
text_a = line
|
| 182 |
+
else:
|
| 183 |
+
text_a = m.group(1)
|
| 184 |
+
text_b = m.group(2)
|
| 185 |
+
examples.append(
|
| 186 |
+
InputExample(unique_id=unique_id, text_a=text_a, text_b=text_b))
|
| 187 |
+
unique_id += 1
|
| 188 |
+
return examples
|
| 189 |
+
|
| 190 |
+
|
| 191 |
+
def main():
|
| 192 |
+
parser = argparse.ArgumentParser()
|
| 193 |
+
|
| 194 |
+
## Required parameters
|
| 195 |
+
parser.add_argument("--input_file", default=None, type=str, required=True)
|
| 196 |
+
parser.add_argument("--output_file", default=None, type=str, required=True)
|
| 197 |
+
parser.add_argument("--bert_model", default=None, type=str, required=True,
|
| 198 |
+
help="Bert pre-trained model selected in the list: bert-base-uncased, "
|
| 199 |
+
"bert-large-uncased, bert-base-cased, bert-base-multilingual, bert-base-chinese.")
|
| 200 |
+
|
| 201 |
+
## Other parameters
|
| 202 |
+
parser.add_argument("--do_lower_case", action='store_true', help="Set this flag if you are using an uncased model.")
|
| 203 |
+
parser.add_argument("--layers", default="-1,-2,-3,-4", type=str)
|
| 204 |
+
parser.add_argument("--max_seq_length", default=128, type=int,
|
| 205 |
+
help="The maximum total input sequence length after WordPiece tokenization. Sequences longer "
|
| 206 |
+
"than this will be truncated, and sequences shorter than this will be padded.")
|
| 207 |
+
parser.add_argument("--batch_size", default=32, type=int, help="Batch size for predictions.")
|
| 208 |
+
parser.add_argument("--local_rank",
|
| 209 |
+
type=int,
|
| 210 |
+
default=-1,
|
| 211 |
+
help = "local_rank for distributed training on gpus")
|
| 212 |
+
parser.add_argument("--no_cuda",
|
| 213 |
+
action='store_true',
|
| 214 |
+
help="Whether not to use CUDA when available")
|
| 215 |
+
|
| 216 |
+
args = parser.parse_args()
|
| 217 |
+
|
| 218 |
+
if args.local_rank == -1 or args.no_cuda:
|
| 219 |
+
device = torch.device("cuda" if torch.cuda.is_available() and not args.no_cuda else "cpu")
|
| 220 |
+
n_gpu = torch.cuda.device_count()
|
| 221 |
+
else:
|
| 222 |
+
device = torch.device("cuda", args.local_rank)
|
| 223 |
+
n_gpu = 1
|
| 224 |
+
# Initializes the distributed backend which will take care of sychronizing nodes/GPUs
|
| 225 |
+
torch.distributed.init_process_group(backend='nccl')
|
| 226 |
+
logger.info("device: {} n_gpu: {} distributed training: {}".format(device, n_gpu, bool(args.local_rank != -1)))
|
| 227 |
+
|
| 228 |
+
layer_indexes = [int(x) for x in args.layers.split(",")]
|
| 229 |
+
|
| 230 |
+
tokenizer = BertTokenizer.from_pretrained(args.bert_model, do_lower_case=args.do_lower_case)
|
| 231 |
+
|
| 232 |
+
examples = read_examples(args.input_file)
|
| 233 |
+
|
| 234 |
+
features = convert_examples_to_features(
|
| 235 |
+
examples=examples, seq_length=args.max_seq_length, tokenizer=tokenizer)
|
| 236 |
+
|
| 237 |
+
unique_id_to_feature = {}
|
| 238 |
+
for feature in features:
|
| 239 |
+
unique_id_to_feature[feature.unique_id] = feature
|
| 240 |
+
|
| 241 |
+
model = BertModel.from_pretrained(args.bert_model)
|
| 242 |
+
model.to(device)
|
| 243 |
+
|
| 244 |
+
if args.local_rank != -1:
|
| 245 |
+
model = torch.nn.parallel.DistributedDataParallel(model, device_ids=[args.local_rank],
|
| 246 |
+
output_device=args.local_rank)
|
| 247 |
+
elif n_gpu > 1:
|
| 248 |
+
model = torch.nn.DataParallel(model)
|
| 249 |
+
|
| 250 |
+
all_input_ids = torch.tensor([f.input_ids for f in features], dtype=torch.long)
|
| 251 |
+
all_input_mask = torch.tensor([f.input_mask for f in features], dtype=torch.long)
|
| 252 |
+
all_example_index = torch.arange(all_input_ids.size(0), dtype=torch.long)
|
| 253 |
+
|
| 254 |
+
eval_data = TensorDataset(all_input_ids, all_input_mask, all_example_index)
|
| 255 |
+
if args.local_rank == -1:
|
| 256 |
+
eval_sampler = SequentialSampler(eval_data)
|
| 257 |
+
else:
|
| 258 |
+
eval_sampler = DistributedSampler(eval_data)
|
| 259 |
+
eval_dataloader = DataLoader(eval_data, sampler=eval_sampler, batch_size=args.batch_size)
|
| 260 |
+
|
| 261 |
+
model.eval()
|
| 262 |
+
with open(args.output_file, "w", encoding='utf-8') as writer:
|
| 263 |
+
for input_ids, input_mask, example_indices in eval_dataloader:
|
| 264 |
+
input_ids = input_ids.to(device)
|
| 265 |
+
input_mask = input_mask.to(device)
|
| 266 |
+
|
| 267 |
+
all_encoder_layers, _ = model(input_ids, token_type_ids=None, attention_mask=input_mask)
|
| 268 |
+
all_encoder_layers = all_encoder_layers
|
| 269 |
+
|
| 270 |
+
for b, example_index in enumerate(example_indices):
|
| 271 |
+
feature = features[example_index.item()]
|
| 272 |
+
unique_id = int(feature.unique_id)
|
| 273 |
+
# feature = unique_id_to_feature[unique_id]
|
| 274 |
+
output_json = collections.OrderedDict()
|
| 275 |
+
output_json["linex_index"] = unique_id
|
| 276 |
+
all_out_features = []
|
| 277 |
+
for (i, token) in enumerate(feature.tokens):
|
| 278 |
+
all_layers = []
|
| 279 |
+
for (j, layer_index) in enumerate(layer_indexes):
|
| 280 |
+
layer_output = all_encoder_layers[int(layer_index)].detach().cpu().numpy()
|
| 281 |
+
layer_output = layer_output[b]
|
| 282 |
+
layers = collections.OrderedDict()
|
| 283 |
+
layers["index"] = layer_index
|
| 284 |
+
layers["values"] = [
|
| 285 |
+
round(x.item(), 6) for x in layer_output[i]
|
| 286 |
+
]
|
| 287 |
+
all_layers.append(layers)
|
| 288 |
+
out_features = collections.OrderedDict()
|
| 289 |
+
out_features["token"] = token
|
| 290 |
+
out_features["layers"] = all_layers
|
| 291 |
+
all_out_features.append(out_features)
|
| 292 |
+
output_json["features"] = all_out_features
|
| 293 |
+
writer.write(json.dumps(output_json) + "\n")
|
| 294 |
+
|
| 295 |
+
|
| 296 |
+
if __name__ == "__main__":
|
| 297 |
+
main()
|
dark-secrets-of-BERT-master/dark-secrets-of-BERT-master/examples/lm_finetuning/README.md
ADDED
|
@@ -0,0 +1,64 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# BERT Model Finetuning using Masked Language Modeling objective
|
| 2 |
+
|
| 3 |
+
## Introduction
|
| 4 |
+
|
| 5 |
+
The three example scripts in this folder can be used to **fine-tune** a pre-trained BERT model using the pretraining objective (combination of masked language modeling and next sentence prediction loss). In general, pretrained models like BERT are first trained with a pretraining objective (masked language modeling and next sentence prediction for BERT) on a large and general natural language corpus. A classifier head is then added on top of the pre-trained architecture and the model is quickly fine-tuned on a target task, while still (hopefully) retaining its general language understanding. This greatly reduces overfitting and yields state-of-the-art results, especially when training data for the target task are limited.
|
| 6 |
+
|
| 7 |
+
The [ULMFiT paper](https://arxiv.org/abs/1801.06146) took a slightly different approach, however, and added an intermediate step in which the model is fine-tuned on text **from the same domain as the target task and using the pretraining objective** before the final stage in which the classifier head is added and the model is trained on the target task itself. This paper reported significantly improved results from this step, and found that they could get high-quality classifications even with only tiny numbers (<1000) of labelled training examples, as long as they had a lot of unlabelled data from the target domain.
|
| 8 |
+
|
| 9 |
+
The BERT model has more capacity than the LSTM models used in the ULMFiT work, but the [BERT paper](https://arxiv.org/abs/1810.04805) did not test finetuning using the pretraining objective and at the present stage there aren't many examples of this approach being used for Transformer-based language models. As such, it's hard to predict what effect this step will have on final model performance, but it's reasonable to conjecture that this approach can improve the final classification performance, especially when a large unlabelled corpus from the target domain is available, labelled data is limited, or the target domain is very unusual and different from 'normal' English text. If you are aware of any literature on this subject, please feel free to add it in here, or open an issue and tag me (@Rocketknight1) and I'll include it.
|
| 10 |
+
|
| 11 |
+
## Input format
|
| 12 |
+
|
| 13 |
+
The scripts in this folder expect a single file as input, consisting of untokenized text, with one **sentence** per line, and one blank line between documents. The reason for the sentence splitting is that part of BERT's training involves a _next sentence_ objective in which the model must predict whether two sequences of text are contiguous text from the same document or not, and to avoid making the task _too easy_, the split point between the sequences is always at the end of a sentence. The linebreaks in the file are therefore necessary to mark the points where the text can be split.
|
| 14 |
+
|
| 15 |
+
## Usage
|
| 16 |
+
|
| 17 |
+
There are two ways to fine-tune a language model using these scripts. The first _quick_ approach is to use [`simple_lm_finetuning.py`](./simple_lm_finetuning.py). This script does everything in a single script, but generates training instances that consist of just two sentences. This is quite different from the BERT paper, where (confusingly) the NextSentence task concatenated sentences together from each document to form two long multi-sentences, which the paper just referred to as _sentences_. The difference between this simple approach and the original paper approach can have a significant effect for long sequences since two sentences will be much shorter than the max sequence length. In this case, most of each training example will just consist of blank padding characters, which wastes a lot of computation and results in a model that isn't really training on long sequences.
|
| 18 |
+
|
| 19 |
+
As such, the preferred approach (assuming you have documents containing multiple contiguous sentences from your target domain) is to use [`pregenerate_training_data.py`](./pregenerate_training_data.py) to pre-process your data into training examples following the methodology used for LM training in the original BERT paper and repository. Since there is a significant random component to training data generation for BERT, this script includes an option to generate multiple _epochs_ of pre-processed data, to avoid training on the same random splits each epoch. Generating an epoch of data for each training epoch should result a better final model, and so we recommend doing so.
|
| 20 |
+
|
| 21 |
+
You can then train on the pregenerated data using [`finetune_on_pregenerated.py`](./finetune_on_pregenerated.py), and pointing it to the folder created by [`pregenerate_training_data.py`](./pregenerate_training_data.py). Note that you should use the same `bert_model` and case options for both! Also note that `max_seq_len` does not need to be specified for the [`finetune_on_pregenerated.py`](./finetune_on_pregenerated.py) script, as it is inferred from the training examples.
|
| 22 |
+
|
| 23 |
+
There are various options that can be tweaked, but they are mostly set to the values from the BERT paper/repository and default values should make sense. The most relevant ones are:
|
| 24 |
+
|
| 25 |
+
- `--max_seq_len`: Controls the length of training examples (in wordpiece tokens) seen by the model. Defaults to 128 but can be set as high as 512. Higher values may yield stronger language models at the cost of slower and more memory-intensive training.
|
| 26 |
+
- `--fp16`: Enables fast half-precision training on recent GPUs.
|
| 27 |
+
|
| 28 |
+
In addition, if memory usage is an issue, especially when training on a single GPU, reducing `--train_batch_size` from the default 32 to a lower number (4-16) can be helpful, or leaving `--train_batch_size` at the default and increasing `--gradient_accumulation_steps` to 2-8. Changing `--gradient_accumulation_steps` may be preferable as alterations to the batch size may require corresponding changes in the learning rate to compensate. There is also a `--reduce_memory` option for both the `pregenerate_training_data.py` and `finetune_on_pregenerated.py` scripts that spills data to disc in shelf objects or numpy memmaps rather than retaining it in memory, which significantly reduces memory usage with little performance impact.
|
| 29 |
+
|
| 30 |
+
## Examples
|
| 31 |
+
|
| 32 |
+
### Simple fine-tuning
|
| 33 |
+
|
| 34 |
+
```
|
| 35 |
+
python3 simple_lm_finetuning.py
|
| 36 |
+
--train_corpus my_corpus.txt
|
| 37 |
+
--bert_model bert-base-uncased
|
| 38 |
+
--do_lower_case
|
| 39 |
+
--output_dir finetuned_lm/
|
| 40 |
+
--do_train
|
| 41 |
+
```
|
| 42 |
+
|
| 43 |
+
### Pregenerating training data
|
| 44 |
+
|
| 45 |
+
```
|
| 46 |
+
python3 pregenerate_training_data.py
|
| 47 |
+
--train_corpus my_corpus.txt
|
| 48 |
+
--bert_model bert-base-uncased
|
| 49 |
+
--do_lower_case
|
| 50 |
+
--output_dir training/
|
| 51 |
+
--epochs_to_generate 3
|
| 52 |
+
--max_seq_len 256
|
| 53 |
+
```
|
| 54 |
+
|
| 55 |
+
### Training on pregenerated data
|
| 56 |
+
|
| 57 |
+
```
|
| 58 |
+
python3 finetune_on_pregenerated.py
|
| 59 |
+
--pregenerated_data training/
|
| 60 |
+
--bert_model bert-base-uncased
|
| 61 |
+
--do_lower_case
|
| 62 |
+
--output_dir finetuned_lm/
|
| 63 |
+
--epochs 3
|
| 64 |
+
```
|
dark-secrets-of-BERT-master/dark-secrets-of-BERT-master/examples/lm_finetuning/finetune_on_pregenerated.py
ADDED
|
@@ -0,0 +1,333 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
from argparse import ArgumentParser
|
| 2 |
+
from pathlib import Path
|
| 3 |
+
import torch
|
| 4 |
+
import logging
|
| 5 |
+
import json
|
| 6 |
+
import random
|
| 7 |
+
import numpy as np
|
| 8 |
+
from collections import namedtuple
|
| 9 |
+
from tempfile import TemporaryDirectory
|
| 10 |
+
|
| 11 |
+
from torch.utils.data import DataLoader, Dataset, RandomSampler
|
| 12 |
+
from torch.utils.data.distributed import DistributedSampler
|
| 13 |
+
from tqdm import tqdm
|
| 14 |
+
|
| 15 |
+
from pytorch_pretrained_bert.modeling import BertForPreTraining
|
| 16 |
+
from pytorch_pretrained_bert.tokenization import BertTokenizer
|
| 17 |
+
from pytorch_pretrained_bert.optimization import BertAdam, warmup_linear
|
| 18 |
+
|
| 19 |
+
InputFeatures = namedtuple("InputFeatures", "input_ids input_mask segment_ids lm_label_ids is_next")
|
| 20 |
+
|
| 21 |
+
log_format = '%(asctime)-10s: %(message)s'
|
| 22 |
+
logging.basicConfig(level=logging.INFO, format=log_format)
|
| 23 |
+
|
| 24 |
+
|
| 25 |
+
def convert_example_to_features(example, tokenizer, max_seq_length):
|
| 26 |
+
tokens = example["tokens"]
|
| 27 |
+
segment_ids = example["segment_ids"]
|
| 28 |
+
is_random_next = example["is_random_next"]
|
| 29 |
+
masked_lm_positions = example["masked_lm_positions"]
|
| 30 |
+
masked_lm_labels = example["masked_lm_labels"]
|
| 31 |
+
|
| 32 |
+
assert len(tokens) == len(segment_ids) <= max_seq_length # The preprocessed data should be already truncated
|
| 33 |
+
input_ids = tokenizer.convert_tokens_to_ids(tokens)
|
| 34 |
+
masked_label_ids = tokenizer.convert_tokens_to_ids(masked_lm_labels)
|
| 35 |
+
|
| 36 |
+
input_array = np.zeros(max_seq_length, dtype=np.int)
|
| 37 |
+
input_array[:len(input_ids)] = input_ids
|
| 38 |
+
|
| 39 |
+
mask_array = np.zeros(max_seq_length, dtype=np.bool)
|
| 40 |
+
mask_array[:len(input_ids)] = 1
|
| 41 |
+
|
| 42 |
+
segment_array = np.zeros(max_seq_length, dtype=np.bool)
|
| 43 |
+
segment_array[:len(segment_ids)] = segment_ids
|
| 44 |
+
|
| 45 |
+
lm_label_array = np.full(max_seq_length, dtype=np.int, fill_value=-1)
|
| 46 |
+
lm_label_array[masked_lm_positions] = masked_label_ids
|
| 47 |
+
|
| 48 |
+
features = InputFeatures(input_ids=input_array,
|
| 49 |
+
input_mask=mask_array,
|
| 50 |
+
segment_ids=segment_array,
|
| 51 |
+
lm_label_ids=lm_label_array,
|
| 52 |
+
is_next=is_random_next)
|
| 53 |
+
return features
|
| 54 |
+
|
| 55 |
+
|
| 56 |
+
class PregeneratedDataset(Dataset):
|
| 57 |
+
def __init__(self, training_path, epoch, tokenizer, num_data_epochs, reduce_memory=False):
|
| 58 |
+
self.vocab = tokenizer.vocab
|
| 59 |
+
self.tokenizer = tokenizer
|
| 60 |
+
self.epoch = epoch
|
| 61 |
+
self.data_epoch = epoch % num_data_epochs
|
| 62 |
+
data_file = training_path / f"epoch_{self.data_epoch}.json"
|
| 63 |
+
metrics_file = training_path / f"epoch_{self.data_epoch}_metrics.json"
|
| 64 |
+
assert data_file.is_file() and metrics_file.is_file()
|
| 65 |
+
metrics = json.loads(metrics_file.read_text())
|
| 66 |
+
num_samples = metrics['num_training_examples']
|
| 67 |
+
seq_len = metrics['max_seq_len']
|
| 68 |
+
self.temp_dir = None
|
| 69 |
+
self.working_dir = None
|
| 70 |
+
if reduce_memory:
|
| 71 |
+
self.temp_dir = TemporaryDirectory()
|
| 72 |
+
self.working_dir = Path(self.temp_dir.name)
|
| 73 |
+
input_ids = np.memmap(filename=self.working_dir/'input_ids.memmap',
|
| 74 |
+
mode='w+', dtype=np.int32, shape=(num_samples, seq_len))
|
| 75 |
+
input_masks = np.memmap(filename=self.working_dir/'input_masks.memmap',
|
| 76 |
+
shape=(num_samples, seq_len), mode='w+', dtype=np.bool)
|
| 77 |
+
segment_ids = np.memmap(filename=self.working_dir/'input_masks.memmap',
|
| 78 |
+
shape=(num_samples, seq_len), mode='w+', dtype=np.bool)
|
| 79 |
+
lm_label_ids = np.memmap(filename=self.working_dir/'lm_label_ids.memmap',
|
| 80 |
+
shape=(num_samples, seq_len), mode='w+', dtype=np.int32)
|
| 81 |
+
lm_label_ids[:] = -1
|
| 82 |
+
is_nexts = np.memmap(filename=self.working_dir/'is_nexts.memmap',
|
| 83 |
+
shape=(num_samples,), mode='w+', dtype=np.bool)
|
| 84 |
+
else:
|
| 85 |
+
input_ids = np.zeros(shape=(num_samples, seq_len), dtype=np.int32)
|
| 86 |
+
input_masks = np.zeros(shape=(num_samples, seq_len), dtype=np.bool)
|
| 87 |
+
segment_ids = np.zeros(shape=(num_samples, seq_len), dtype=np.bool)
|
| 88 |
+
lm_label_ids = np.full(shape=(num_samples, seq_len), dtype=np.int32, fill_value=-1)
|
| 89 |
+
is_nexts = np.zeros(shape=(num_samples,), dtype=np.bool)
|
| 90 |
+
logging.info(f"Loading training examples for epoch {epoch}")
|
| 91 |
+
with data_file.open() as f:
|
| 92 |
+
for i, line in enumerate(tqdm(f, total=num_samples, desc="Training examples")):
|
| 93 |
+
line = line.strip()
|
| 94 |
+
example = json.loads(line)
|
| 95 |
+
features = convert_example_to_features(example, tokenizer, seq_len)
|
| 96 |
+
input_ids[i] = features.input_ids
|
| 97 |
+
segment_ids[i] = features.segment_ids
|
| 98 |
+
input_masks[i] = features.input_mask
|
| 99 |
+
lm_label_ids[i] = features.lm_label_ids
|
| 100 |
+
is_nexts[i] = features.is_next
|
| 101 |
+
assert i == num_samples - 1 # Assert that the sample count metric was true
|
| 102 |
+
logging.info("Loading complete!")
|
| 103 |
+
self.num_samples = num_samples
|
| 104 |
+
self.seq_len = seq_len
|
| 105 |
+
self.input_ids = input_ids
|
| 106 |
+
self.input_masks = input_masks
|
| 107 |
+
self.segment_ids = segment_ids
|
| 108 |
+
self.lm_label_ids = lm_label_ids
|
| 109 |
+
self.is_nexts = is_nexts
|
| 110 |
+
|
| 111 |
+
def __len__(self):
|
| 112 |
+
return self.num_samples
|
| 113 |
+
|
| 114 |
+
def __getitem__(self, item):
|
| 115 |
+
return (torch.tensor(self.input_ids[item].astype(np.int64)),
|
| 116 |
+
torch.tensor(self.input_masks[item].astype(np.int64)),
|
| 117 |
+
torch.tensor(self.segment_ids[item].astype(np.int64)),
|
| 118 |
+
torch.tensor(self.lm_label_ids[item].astype(np.int64)),
|
| 119 |
+
torch.tensor(self.is_nexts[item].astype(np.int64)))
|
| 120 |
+
|
| 121 |
+
|
| 122 |
+
def main():
|
| 123 |
+
parser = ArgumentParser()
|
| 124 |
+
parser.add_argument('--pregenerated_data', type=Path, required=True)
|
| 125 |
+
parser.add_argument('--output_dir', type=Path, required=True)
|
| 126 |
+
parser.add_argument("--bert_model", type=str, required=True, help="Bert pre-trained model selected in the list: bert-base-uncased, "
|
| 127 |
+
"bert-large-uncased, bert-base-cased, bert-base-multilingual, bert-base-chinese.")
|
| 128 |
+
parser.add_argument("--do_lower_case", action="store_true")
|
| 129 |
+
parser.add_argument("--reduce_memory", action="store_true",
|
| 130 |
+
help="Store training data as on-disc memmaps to massively reduce memory usage")
|
| 131 |
+
|
| 132 |
+
parser.add_argument("--epochs", type=int, default=3, help="Number of epochs to train for")
|
| 133 |
+
parser.add_argument("--local_rank",
|
| 134 |
+
type=int,
|
| 135 |
+
default=-1,
|
| 136 |
+
help="local_rank for distributed training on gpus")
|
| 137 |
+
parser.add_argument("--no_cuda",
|
| 138 |
+
action='store_true',
|
| 139 |
+
help="Whether not to use CUDA when available")
|
| 140 |
+
parser.add_argument('--gradient_accumulation_steps',
|
| 141 |
+
type=int,
|
| 142 |
+
default=1,
|
| 143 |
+
help="Number of updates steps to accumulate before performing a backward/update pass.")
|
| 144 |
+
parser.add_argument("--train_batch_size",
|
| 145 |
+
default=32,
|
| 146 |
+
type=int,
|
| 147 |
+
help="Total batch size for training.")
|
| 148 |
+
parser.add_argument('--fp16',
|
| 149 |
+
action='store_true',
|
| 150 |
+
help="Whether to use 16-bit float precision instead of 32-bit")
|
| 151 |
+
parser.add_argument('--loss_scale',
|
| 152 |
+
type=float, default=0,
|
| 153 |
+
help="Loss scaling to improve fp16 numeric stability. Only used when fp16 set to True.\n"
|
| 154 |
+
"0 (default value): dynamic loss scaling.\n"
|
| 155 |
+
"Positive power of 2: static loss scaling value.\n")
|
| 156 |
+
parser.add_argument("--warmup_proportion",
|
| 157 |
+
default=0.1,
|
| 158 |
+
type=float,
|
| 159 |
+
help="Proportion of training to perform linear learning rate warmup for. "
|
| 160 |
+
"E.g., 0.1 = 10%% of training.")
|
| 161 |
+
parser.add_argument("--learning_rate",
|
| 162 |
+
default=3e-5,
|
| 163 |
+
type=float,
|
| 164 |
+
help="The initial learning rate for Adam.")
|
| 165 |
+
parser.add_argument('--seed',
|
| 166 |
+
type=int,
|
| 167 |
+
default=42,
|
| 168 |
+
help="random seed for initialization")
|
| 169 |
+
args = parser.parse_args()
|
| 170 |
+
|
| 171 |
+
assert args.pregenerated_data.is_dir(), \
|
| 172 |
+
"--pregenerated_data should point to the folder of files made by pregenerate_training_data.py!"
|
| 173 |
+
|
| 174 |
+
samples_per_epoch = []
|
| 175 |
+
for i in range(args.epochs):
|
| 176 |
+
epoch_file = args.pregenerated_data / f"epoch_{i}.json"
|
| 177 |
+
metrics_file = args.pregenerated_data / f"epoch_{i}_metrics.json"
|
| 178 |
+
if epoch_file.is_file() and metrics_file.is_file():
|
| 179 |
+
metrics = json.loads(metrics_file.read_text())
|
| 180 |
+
samples_per_epoch.append(metrics['num_training_examples'])
|
| 181 |
+
else:
|
| 182 |
+
if i == 0:
|
| 183 |
+
exit("No training data was found!")
|
| 184 |
+
print(f"Warning! There are fewer epochs of pregenerated data ({i}) than training epochs ({args.epochs}).")
|
| 185 |
+
print("This script will loop over the available data, but training diversity may be negatively impacted.")
|
| 186 |
+
num_data_epochs = i
|
| 187 |
+
break
|
| 188 |
+
else:
|
| 189 |
+
num_data_epochs = args.epochs
|
| 190 |
+
|
| 191 |
+
if args.local_rank == -1 or args.no_cuda:
|
| 192 |
+
device = torch.device("cuda" if torch.cuda.is_available() and not args.no_cuda else "cpu")
|
| 193 |
+
n_gpu = torch.cuda.device_count()
|
| 194 |
+
else:
|
| 195 |
+
torch.cuda.set_device(args.local_rank)
|
| 196 |
+
device = torch.device("cuda", args.local_rank)
|
| 197 |
+
n_gpu = 1
|
| 198 |
+
# Initializes the distributed backend which will take care of sychronizing nodes/GPUs
|
| 199 |
+
torch.distributed.init_process_group(backend='nccl')
|
| 200 |
+
logging.info("device: {} n_gpu: {}, distributed training: {}, 16-bits training: {}".format(
|
| 201 |
+
device, n_gpu, bool(args.local_rank != -1), args.fp16))
|
| 202 |
+
|
| 203 |
+
if args.gradient_accumulation_steps < 1:
|
| 204 |
+
raise ValueError("Invalid gradient_accumulation_steps parameter: {}, should be >= 1".format(
|
| 205 |
+
args.gradient_accumulation_steps))
|
| 206 |
+
|
| 207 |
+
args.train_batch_size = args.train_batch_size // args.gradient_accumulation_steps
|
| 208 |
+
|
| 209 |
+
random.seed(args.seed)
|
| 210 |
+
np.random.seed(args.seed)
|
| 211 |
+
torch.manual_seed(args.seed)
|
| 212 |
+
if n_gpu > 0:
|
| 213 |
+
torch.cuda.manual_seed_all(args.seed)
|
| 214 |
+
|
| 215 |
+
if args.output_dir.is_dir() and list(args.output_dir.iterdir()):
|
| 216 |
+
logging.warning(f"Output directory ({args.output_dir}) already exists and is not empty!")
|
| 217 |
+
args.output_dir.mkdir(parents=True, exist_ok=True)
|
| 218 |
+
|
| 219 |
+
tokenizer = BertTokenizer.from_pretrained(args.bert_model, do_lower_case=args.do_lower_case)
|
| 220 |
+
|
| 221 |
+
total_train_examples = 0
|
| 222 |
+
for i in range(args.epochs):
|
| 223 |
+
# The modulo takes into account the fact that we may loop over limited epochs of data
|
| 224 |
+
total_train_examples += samples_per_epoch[i % len(samples_per_epoch)]
|
| 225 |
+
|
| 226 |
+
num_train_optimization_steps = int(
|
| 227 |
+
total_train_examples / args.train_batch_size / args.gradient_accumulation_steps)
|
| 228 |
+
if args.local_rank != -1:
|
| 229 |
+
num_train_optimization_steps = num_train_optimization_steps // torch.distributed.get_world_size()
|
| 230 |
+
|
| 231 |
+
# Prepare model
|
| 232 |
+
model = BertForPreTraining.from_pretrained(args.bert_model)
|
| 233 |
+
if args.fp16:
|
| 234 |
+
model.half()
|
| 235 |
+
model.to(device)
|
| 236 |
+
if args.local_rank != -1:
|
| 237 |
+
try:
|
| 238 |
+
from apex.parallel import DistributedDataParallel as DDP
|
| 239 |
+
except ImportError:
|
| 240 |
+
raise ImportError(
|
| 241 |
+
"Please install apex from https://www.github.com/nvidia/apex to use distributed and fp16 training.")
|
| 242 |
+
model = DDP(model)
|
| 243 |
+
elif n_gpu > 1:
|
| 244 |
+
model = torch.nn.DataParallel(model)
|
| 245 |
+
|
| 246 |
+
# Prepare optimizer
|
| 247 |
+
param_optimizer = list(model.named_parameters())
|
| 248 |
+
no_decay = ['bias', 'LayerNorm.bias', 'LayerNorm.weight']
|
| 249 |
+
optimizer_grouped_parameters = [
|
| 250 |
+
{'params': [p for n, p in param_optimizer if not any(nd in n for nd in no_decay)],
|
| 251 |
+
'weight_decay': 0.01},
|
| 252 |
+
{'params': [p for n, p in param_optimizer if any(nd in n for nd in no_decay)], 'weight_decay': 0.0}
|
| 253 |
+
]
|
| 254 |
+
|
| 255 |
+
if args.fp16:
|
| 256 |
+
try:
|
| 257 |
+
from apex.optimizers import FP16_Optimizer
|
| 258 |
+
from apex.optimizers import FusedAdam
|
| 259 |
+
except ImportError:
|
| 260 |
+
raise ImportError(
|
| 261 |
+
"Please install apex from https://www.github.com/nvidia/apex to use distributed and fp16 training.")
|
| 262 |
+
|
| 263 |
+
optimizer = FusedAdam(optimizer_grouped_parameters,
|
| 264 |
+
lr=args.learning_rate,
|
| 265 |
+
bias_correction=False,
|
| 266 |
+
max_grad_norm=1.0)
|
| 267 |
+
if args.loss_scale == 0:
|
| 268 |
+
optimizer = FP16_Optimizer(optimizer, dynamic_loss_scale=True)
|
| 269 |
+
else:
|
| 270 |
+
optimizer = FP16_Optimizer(optimizer, static_loss_scale=args.loss_scale)
|
| 271 |
+
|
| 272 |
+
else:
|
| 273 |
+
optimizer = BertAdam(optimizer_grouped_parameters,
|
| 274 |
+
lr=args.learning_rate,
|
| 275 |
+
warmup=args.warmup_proportion,
|
| 276 |
+
t_total=num_train_optimization_steps)
|
| 277 |
+
|
| 278 |
+
global_step = 0
|
| 279 |
+
logging.info("***** Running training *****")
|
| 280 |
+
logging.info(f" Num examples = {total_train_examples}")
|
| 281 |
+
logging.info(" Batch size = %d", args.train_batch_size)
|
| 282 |
+
logging.info(" Num steps = %d", num_train_optimization_steps)
|
| 283 |
+
model.train()
|
| 284 |
+
for epoch in range(args.epochs):
|
| 285 |
+
epoch_dataset = PregeneratedDataset(epoch=epoch, training_path=args.pregenerated_data, tokenizer=tokenizer,
|
| 286 |
+
num_data_epochs=num_data_epochs)
|
| 287 |
+
if args.local_rank == -1:
|
| 288 |
+
train_sampler = RandomSampler(epoch_dataset)
|
| 289 |
+
else:
|
| 290 |
+
train_sampler = DistributedSampler(epoch_dataset)
|
| 291 |
+
train_dataloader = DataLoader(epoch_dataset, sampler=train_sampler, batch_size=args.train_batch_size)
|
| 292 |
+
tr_loss = 0
|
| 293 |
+
nb_tr_examples, nb_tr_steps = 0, 0
|
| 294 |
+
with tqdm(total=len(train_dataloader), desc=f"Epoch {epoch}") as pbar:
|
| 295 |
+
for step, batch in enumerate(train_dataloader):
|
| 296 |
+
batch = tuple(t.to(device) for t in batch)
|
| 297 |
+
input_ids, input_mask, segment_ids, lm_label_ids, is_next = batch
|
| 298 |
+
loss = model(input_ids, segment_ids, input_mask, lm_label_ids, is_next)
|
| 299 |
+
if n_gpu > 1:
|
| 300 |
+
loss = loss.mean() # mean() to average on multi-gpu.
|
| 301 |
+
if args.gradient_accumulation_steps > 1:
|
| 302 |
+
loss = loss / args.gradient_accumulation_steps
|
| 303 |
+
if args.fp16:
|
| 304 |
+
optimizer.backward(loss)
|
| 305 |
+
else:
|
| 306 |
+
loss.backward()
|
| 307 |
+
tr_loss += loss.item()
|
| 308 |
+
nb_tr_examples += input_ids.size(0)
|
| 309 |
+
nb_tr_steps += 1
|
| 310 |
+
pbar.update(1)
|
| 311 |
+
mean_loss = tr_loss * args.gradient_accumulation_steps / nb_tr_steps
|
| 312 |
+
pbar.set_postfix_str(f"Loss: {mean_loss:.5f}")
|
| 313 |
+
if (step + 1) % args.gradient_accumulation_steps == 0:
|
| 314 |
+
if args.fp16:
|
| 315 |
+
# modify learning rate with special warm up BERT uses
|
| 316 |
+
# if args.fp16 is False, BertAdam is used that handles this automatically
|
| 317 |
+
lr_this_step = args.learning_rate * warmup_linear(global_step/num_train_optimization_steps,
|
| 318 |
+
args.warmup_proportion)
|
| 319 |
+
for param_group in optimizer.param_groups:
|
| 320 |
+
param_group['lr'] = lr_this_step
|
| 321 |
+
optimizer.step()
|
| 322 |
+
optimizer.zero_grad()
|
| 323 |
+
global_step += 1
|
| 324 |
+
|
| 325 |
+
# Save a trained model
|
| 326 |
+
logging.info("** ** * Saving fine-tuned model ** ** * ")
|
| 327 |
+
model_to_save = model.module if hasattr(model, 'module') else model # Only save the model it-self
|
| 328 |
+
output_model_file = args.output_dir / "pytorch_model.bin"
|
| 329 |
+
torch.save(model_to_save.state_dict(), str(output_model_file))
|
| 330 |
+
|
| 331 |
+
|
| 332 |
+
if __name__ == '__main__':
|
| 333 |
+
main()
|
dark-secrets-of-BERT-master/dark-secrets-of-BERT-master/examples/lm_finetuning/pregenerate_training_data.py
ADDED
|
@@ -0,0 +1,302 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
from argparse import ArgumentParser
|
| 2 |
+
from pathlib import Path
|
| 3 |
+
from tqdm import tqdm, trange
|
| 4 |
+
from tempfile import TemporaryDirectory
|
| 5 |
+
import shelve
|
| 6 |
+
|
| 7 |
+
from random import random, randrange, randint, shuffle, choice, sample
|
| 8 |
+
from pytorch_pretrained_bert.tokenization import BertTokenizer
|
| 9 |
+
import numpy as np
|
| 10 |
+
import json
|
| 11 |
+
|
| 12 |
+
|
| 13 |
+
class DocumentDatabase:
|
| 14 |
+
def __init__(self, reduce_memory=False):
|
| 15 |
+
if reduce_memory:
|
| 16 |
+
self.temp_dir = TemporaryDirectory()
|
| 17 |
+
self.working_dir = Path(self.temp_dir.name)
|
| 18 |
+
self.document_shelf_filepath = self.working_dir / 'shelf.db'
|
| 19 |
+
self.document_shelf = shelve.open(str(self.document_shelf_filepath),
|
| 20 |
+
flag='n', protocol=-1)
|
| 21 |
+
self.documents = None
|
| 22 |
+
else:
|
| 23 |
+
self.documents = []
|
| 24 |
+
self.document_shelf = None
|
| 25 |
+
self.document_shelf_filepath = None
|
| 26 |
+
self.temp_dir = None
|
| 27 |
+
self.doc_lengths = []
|
| 28 |
+
self.doc_cumsum = None
|
| 29 |
+
self.cumsum_max = None
|
| 30 |
+
self.reduce_memory = reduce_memory
|
| 31 |
+
|
| 32 |
+
def add_document(self, document):
|
| 33 |
+
if not document:
|
| 34 |
+
return
|
| 35 |
+
if self.reduce_memory:
|
| 36 |
+
current_idx = len(self.doc_lengths)
|
| 37 |
+
self.document_shelf[str(current_idx)] = document
|
| 38 |
+
else:
|
| 39 |
+
self.documents.append(document)
|
| 40 |
+
self.doc_lengths.append(len(document))
|
| 41 |
+
|
| 42 |
+
def _precalculate_doc_weights(self):
|
| 43 |
+
self.doc_cumsum = np.cumsum(self.doc_lengths)
|
| 44 |
+
self.cumsum_max = self.doc_cumsum[-1]
|
| 45 |
+
|
| 46 |
+
def sample_doc(self, current_idx, sentence_weighted=True):
|
| 47 |
+
# Uses the current iteration counter to ensure we don't sample the same doc twice
|
| 48 |
+
if sentence_weighted:
|
| 49 |
+
# With sentence weighting, we sample docs proportionally to their sentence length
|
| 50 |
+
if self.doc_cumsum is None or len(self.doc_cumsum) != len(self.doc_lengths):
|
| 51 |
+
self._precalculate_doc_weights()
|
| 52 |
+
rand_start = self.doc_cumsum[current_idx]
|
| 53 |
+
rand_end = rand_start + self.cumsum_max - self.doc_lengths[current_idx]
|
| 54 |
+
sentence_index = randrange(rand_start, rand_end) % self.cumsum_max
|
| 55 |
+
sampled_doc_index = np.searchsorted(self.doc_cumsum, sentence_index, side='right')
|
| 56 |
+
else:
|
| 57 |
+
# If we don't use sentence weighting, then every doc has an equal chance to be chosen
|
| 58 |
+
sampled_doc_index = (current_idx + randrange(1, len(self.doc_lengths))) % len(self.doc_lengths)
|
| 59 |
+
assert sampled_doc_index != current_idx
|
| 60 |
+
if self.reduce_memory:
|
| 61 |
+
return self.document_shelf[str(sampled_doc_index)]
|
| 62 |
+
else:
|
| 63 |
+
return self.documents[sampled_doc_index]
|
| 64 |
+
|
| 65 |
+
def __len__(self):
|
| 66 |
+
return len(self.doc_lengths)
|
| 67 |
+
|
| 68 |
+
def __getitem__(self, item):
|
| 69 |
+
if self.reduce_memory:
|
| 70 |
+
return self.document_shelf[str(item)]
|
| 71 |
+
else:
|
| 72 |
+
return self.documents[item]
|
| 73 |
+
|
| 74 |
+
def __enter__(self):
|
| 75 |
+
return self
|
| 76 |
+
|
| 77 |
+
def __exit__(self, exc_type, exc_val, traceback):
|
| 78 |
+
if self.document_shelf is not None:
|
| 79 |
+
self.document_shelf.close()
|
| 80 |
+
if self.temp_dir is not None:
|
| 81 |
+
self.temp_dir.cleanup()
|
| 82 |
+
|
| 83 |
+
|
| 84 |
+
def truncate_seq_pair(tokens_a, tokens_b, max_num_tokens):
|
| 85 |
+
"""Truncates a pair of sequences to a maximum sequence length. Lifted from Google's BERT repo."""
|
| 86 |
+
while True:
|
| 87 |
+
total_length = len(tokens_a) + len(tokens_b)
|
| 88 |
+
if total_length <= max_num_tokens:
|
| 89 |
+
break
|
| 90 |
+
|
| 91 |
+
trunc_tokens = tokens_a if len(tokens_a) > len(tokens_b) else tokens_b
|
| 92 |
+
assert len(trunc_tokens) >= 1
|
| 93 |
+
|
| 94 |
+
# We want to sometimes truncate from the front and sometimes from the
|
| 95 |
+
# back to add more randomness and avoid biases.
|
| 96 |
+
if random() < 0.5:
|
| 97 |
+
del trunc_tokens[0]
|
| 98 |
+
else:
|
| 99 |
+
trunc_tokens.pop()
|
| 100 |
+
|
| 101 |
+
|
| 102 |
+
def create_masked_lm_predictions(tokens, masked_lm_prob, max_predictions_per_seq, vocab_list):
|
| 103 |
+
"""Creates the predictions for the masked LM objective. This is mostly copied from the Google BERT repo, but
|
| 104 |
+
with several refactors to clean it up and remove a lot of unnecessary variables."""
|
| 105 |
+
cand_indices = []
|
| 106 |
+
for (i, token) in enumerate(tokens):
|
| 107 |
+
if token == "[CLS]" or token == "[SEP]":
|
| 108 |
+
continue
|
| 109 |
+
cand_indices.append(i)
|
| 110 |
+
|
| 111 |
+
num_to_mask = min(max_predictions_per_seq,
|
| 112 |
+
max(1, int(round(len(tokens) * masked_lm_prob))))
|
| 113 |
+
shuffle(cand_indices)
|
| 114 |
+
mask_indices = sorted(sample(cand_indices, num_to_mask))
|
| 115 |
+
masked_token_labels = []
|
| 116 |
+
for index in mask_indices:
|
| 117 |
+
# 80% of the time, replace with [MASK]
|
| 118 |
+
if random() < 0.8:
|
| 119 |
+
masked_token = "[MASK]"
|
| 120 |
+
else:
|
| 121 |
+
# 10% of the time, keep original
|
| 122 |
+
if random() < 0.5:
|
| 123 |
+
masked_token = tokens[index]
|
| 124 |
+
# 10% of the time, replace with random word
|
| 125 |
+
else:
|
| 126 |
+
masked_token = choice(vocab_list)
|
| 127 |
+
masked_token_labels.append(tokens[index])
|
| 128 |
+
# Once we've saved the true label for that token, we can overwrite it with the masked version
|
| 129 |
+
tokens[index] = masked_token
|
| 130 |
+
|
| 131 |
+
return tokens, mask_indices, masked_token_labels
|
| 132 |
+
|
| 133 |
+
|
| 134 |
+
def create_instances_from_document(
|
| 135 |
+
doc_database, doc_idx, max_seq_length, short_seq_prob,
|
| 136 |
+
masked_lm_prob, max_predictions_per_seq, vocab_list):
|
| 137 |
+
"""This code is mostly a duplicate of the equivalent function from Google BERT's repo.
|
| 138 |
+
However, we make some changes and improvements. Sampling is improved and no longer requires a loop in this function.
|
| 139 |
+
Also, documents are sampled proportionally to the number of sentences they contain, which means each sentence
|
| 140 |
+
(rather than each document) has an equal chance of being sampled as a false example for the NextSentence task."""
|
| 141 |
+
document = doc_database[doc_idx]
|
| 142 |
+
# Account for [CLS], [SEP], [SEP]
|
| 143 |
+
max_num_tokens = max_seq_length - 3
|
| 144 |
+
|
| 145 |
+
# We *usually* want to fill up the entire sequence since we are padding
|
| 146 |
+
# to `max_seq_length` anyways, so short sequences are generally wasted
|
| 147 |
+
# computation. However, we *sometimes*
|
| 148 |
+
# (i.e., short_seq_prob == 0.1 == 10% of the time) want to use shorter
|
| 149 |
+
# sequences to minimize the mismatch between pre-training and fine-tuning.
|
| 150 |
+
# The `target_seq_length` is just a rough target however, whereas
|
| 151 |
+
# `max_seq_length` is a hard limit.
|
| 152 |
+
target_seq_length = max_num_tokens
|
| 153 |
+
if random() < short_seq_prob:
|
| 154 |
+
target_seq_length = randint(2, max_num_tokens)
|
| 155 |
+
|
| 156 |
+
# We DON'T just concatenate all of the tokens from a document into a long
|
| 157 |
+
# sequence and choose an arbitrary split point because this would make the
|
| 158 |
+
# next sentence prediction task too easy. Instead, we split the input into
|
| 159 |
+
# segments "A" and "B" based on the actual "sentences" provided by the user
|
| 160 |
+
# input.
|
| 161 |
+
instances = []
|
| 162 |
+
current_chunk = []
|
| 163 |
+
current_length = 0
|
| 164 |
+
i = 0
|
| 165 |
+
while i < len(document):
|
| 166 |
+
segment = document[i]
|
| 167 |
+
current_chunk.append(segment)
|
| 168 |
+
current_length += len(segment)
|
| 169 |
+
if i == len(document) - 1 or current_length >= target_seq_length:
|
| 170 |
+
if current_chunk:
|
| 171 |
+
# `a_end` is how many segments from `current_chunk` go into the `A`
|
| 172 |
+
# (first) sentence.
|
| 173 |
+
a_end = 1
|
| 174 |
+
if len(current_chunk) >= 2:
|
| 175 |
+
a_end = randrange(1, len(current_chunk))
|
| 176 |
+
|
| 177 |
+
tokens_a = []
|
| 178 |
+
for j in range(a_end):
|
| 179 |
+
tokens_a.extend(current_chunk[j])
|
| 180 |
+
|
| 181 |
+
tokens_b = []
|
| 182 |
+
|
| 183 |
+
# Random next
|
| 184 |
+
if len(current_chunk) == 1 or random() < 0.5:
|
| 185 |
+
is_random_next = True
|
| 186 |
+
target_b_length = target_seq_length - len(tokens_a)
|
| 187 |
+
|
| 188 |
+
# Sample a random document, with longer docs being sampled more frequently
|
| 189 |
+
random_document = doc_database.sample_doc(current_idx=doc_idx, sentence_weighted=True)
|
| 190 |
+
|
| 191 |
+
random_start = randrange(0, len(random_document))
|
| 192 |
+
for j in range(random_start, len(random_document)):
|
| 193 |
+
tokens_b.extend(random_document[j])
|
| 194 |
+
if len(tokens_b) >= target_b_length:
|
| 195 |
+
break
|
| 196 |
+
# We didn't actually use these segments so we "put them back" so
|
| 197 |
+
# they don't go to waste.
|
| 198 |
+
num_unused_segments = len(current_chunk) - a_end
|
| 199 |
+
i -= num_unused_segments
|
| 200 |
+
# Actual next
|
| 201 |
+
else:
|
| 202 |
+
is_random_next = False
|
| 203 |
+
for j in range(a_end, len(current_chunk)):
|
| 204 |
+
tokens_b.extend(current_chunk[j])
|
| 205 |
+
truncate_seq_pair(tokens_a, tokens_b, max_num_tokens)
|
| 206 |
+
|
| 207 |
+
assert len(tokens_a) >= 1
|
| 208 |
+
assert len(tokens_b) >= 1
|
| 209 |
+
|
| 210 |
+
tokens = ["[CLS]"] + tokens_a + ["[SEP]"] + tokens_b + ["[SEP]"]
|
| 211 |
+
# The segment IDs are 0 for the [CLS] token, the A tokens and the first [SEP]
|
| 212 |
+
# They are 1 for the B tokens and the final [SEP]
|
| 213 |
+
segment_ids = [0 for _ in range(len(tokens_a) + 2)] + [1 for _ in range(len(tokens_b) + 1)]
|
| 214 |
+
|
| 215 |
+
tokens, masked_lm_positions, masked_lm_labels = create_masked_lm_predictions(
|
| 216 |
+
tokens, masked_lm_prob, max_predictions_per_seq, vocab_list)
|
| 217 |
+
|
| 218 |
+
instance = {
|
| 219 |
+
"tokens": tokens,
|
| 220 |
+
"segment_ids": segment_ids,
|
| 221 |
+
"is_random_next": is_random_next,
|
| 222 |
+
"masked_lm_positions": masked_lm_positions,
|
| 223 |
+
"masked_lm_labels": masked_lm_labels}
|
| 224 |
+
instances.append(instance)
|
| 225 |
+
current_chunk = []
|
| 226 |
+
current_length = 0
|
| 227 |
+
i += 1
|
| 228 |
+
|
| 229 |
+
return instances
|
| 230 |
+
|
| 231 |
+
|
| 232 |
+
def main():
|
| 233 |
+
parser = ArgumentParser()
|
| 234 |
+
parser.add_argument('--train_corpus', type=Path, required=True)
|
| 235 |
+
parser.add_argument("--output_dir", type=Path, required=True)
|
| 236 |
+
parser.add_argument("--bert_model", type=str, required=True,
|
| 237 |
+
choices=["bert-base-uncased", "bert-large-uncased", "bert-base-cased",
|
| 238 |
+
"bert-base-multilingual", "bert-base-chinese"])
|
| 239 |
+
parser.add_argument("--do_lower_case", action="store_true")
|
| 240 |
+
|
| 241 |
+
parser.add_argument("--reduce_memory", action="store_true",
|
| 242 |
+
help="Reduce memory usage for large datasets by keeping data on disc rather than in memory")
|
| 243 |
+
|
| 244 |
+
parser.add_argument("--epochs_to_generate", type=int, default=3,
|
| 245 |
+
help="Number of epochs of data to pregenerate")
|
| 246 |
+
parser.add_argument("--max_seq_len", type=int, default=128)
|
| 247 |
+
parser.add_argument("--short_seq_prob", type=float, default=0.1,
|
| 248 |
+
help="Probability of making a short sentence as a training example")
|
| 249 |
+
parser.add_argument("--masked_lm_prob", type=float, default=0.15,
|
| 250 |
+
help="Probability of masking each token for the LM task")
|
| 251 |
+
parser.add_argument("--max_predictions_per_seq", type=int, default=20,
|
| 252 |
+
help="Maximum number of tokens to mask in each sequence")
|
| 253 |
+
|
| 254 |
+
args = parser.parse_args()
|
| 255 |
+
|
| 256 |
+
tokenizer = BertTokenizer.from_pretrained(args.bert_model, do_lower_case=args.do_lower_case)
|
| 257 |
+
vocab_list = list(tokenizer.vocab.keys())
|
| 258 |
+
with DocumentDatabase(reduce_memory=args.reduce_memory) as docs:
|
| 259 |
+
with args.train_corpus.open() as f:
|
| 260 |
+
doc = []
|
| 261 |
+
for line in tqdm(f, desc="Loading Dataset", unit=" lines"):
|
| 262 |
+
line = line.strip()
|
| 263 |
+
if line == "":
|
| 264 |
+
docs.add_document(doc)
|
| 265 |
+
doc = []
|
| 266 |
+
else:
|
| 267 |
+
tokens = tokenizer.tokenize(line)
|
| 268 |
+
doc.append(tokens)
|
| 269 |
+
if doc:
|
| 270 |
+
docs.add_document(doc) # If the last doc didn't end on a newline, make sure it still gets added
|
| 271 |
+
if len(docs) <= 1:
|
| 272 |
+
exit("ERROR: No document breaks were found in the input file! These are necessary to allow the script to "
|
| 273 |
+
"ensure that random NextSentences are not sampled from the same document. Please add blank lines to "
|
| 274 |
+
"indicate breaks between documents in your input file. If your dataset does not contain multiple "
|
| 275 |
+
"documents, blank lines can be inserted at any natural boundary, such as the ends of chapters, "
|
| 276 |
+
"sections or paragraphs.")
|
| 277 |
+
|
| 278 |
+
args.output_dir.mkdir(exist_ok=True)
|
| 279 |
+
for epoch in trange(args.epochs_to_generate, desc="Epoch"):
|
| 280 |
+
epoch_filename = args.output_dir / f"epoch_{epoch}.json"
|
| 281 |
+
num_instances = 0
|
| 282 |
+
with epoch_filename.open('w') as epoch_file:
|
| 283 |
+
for doc_idx in trange(len(docs), desc="Document"):
|
| 284 |
+
doc_instances = create_instances_from_document(
|
| 285 |
+
docs, doc_idx, max_seq_length=args.max_seq_len, short_seq_prob=args.short_seq_prob,
|
| 286 |
+
masked_lm_prob=args.masked_lm_prob, max_predictions_per_seq=args.max_predictions_per_seq,
|
| 287 |
+
vocab_list=vocab_list)
|
| 288 |
+
doc_instances = [json.dumps(instance) for instance in doc_instances]
|
| 289 |
+
for instance in doc_instances:
|
| 290 |
+
epoch_file.write(instance + '\n')
|
| 291 |
+
num_instances += 1
|
| 292 |
+
metrics_file = args.output_dir / f"epoch_{epoch}_metrics.json"
|
| 293 |
+
with metrics_file.open('w') as metrics_file:
|
| 294 |
+
metrics = {
|
| 295 |
+
"num_training_examples": num_instances,
|
| 296 |
+
"max_seq_len": args.max_seq_len
|
| 297 |
+
}
|
| 298 |
+
metrics_file.write(json.dumps(metrics))
|
| 299 |
+
|
| 300 |
+
|
| 301 |
+
if __name__ == '__main__':
|
| 302 |
+
main()
|
dark-secrets-of-BERT-master/dark-secrets-of-BERT-master/examples/lm_finetuning/simple_lm_finetuning.py
ADDED
|
@@ -0,0 +1,642 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# coding=utf-8
|
| 2 |
+
# Copyright 2018 The Google AI Language Team Authors and The HuggingFace Inc. team.
|
| 3 |
+
# Copyright (c) 2018, NVIDIA CORPORATION. All rights reserved.
|
| 4 |
+
#
|
| 5 |
+
# Licensed under the Apache License, Version 2.0 (the "License");
|
| 6 |
+
# you may not use this file except in compliance with the License.
|
| 7 |
+
# You may obtain a copy of the License at
|
| 8 |
+
#
|
| 9 |
+
# http://www.apache.org/licenses/LICENSE-2.0
|
| 10 |
+
#
|
| 11 |
+
# Unless required by applicable law or agreed to in writing, software
|
| 12 |
+
# distributed under the License is distributed on an "AS IS" BASIS,
|
| 13 |
+
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
| 14 |
+
# See the License for the specific language governing permissions and
|
| 15 |
+
# limitations under the License.
|
| 16 |
+
"""BERT finetuning runner."""
|
| 17 |
+
|
| 18 |
+
from __future__ import absolute_import, division, print_function, unicode_literals
|
| 19 |
+
|
| 20 |
+
import argparse
|
| 21 |
+
import logging
|
| 22 |
+
import os
|
| 23 |
+
import random
|
| 24 |
+
from io import open
|
| 25 |
+
|
| 26 |
+
import numpy as np
|
| 27 |
+
import torch
|
| 28 |
+
from torch.utils.data import DataLoader, Dataset, RandomSampler
|
| 29 |
+
from torch.utils.data.distributed import DistributedSampler
|
| 30 |
+
from tqdm import tqdm, trange
|
| 31 |
+
|
| 32 |
+
from pytorch_pretrained_bert.modeling import BertForPreTraining
|
| 33 |
+
from pytorch_pretrained_bert.tokenization import BertTokenizer
|
| 34 |
+
from pytorch_pretrained_bert.optimization import BertAdam, warmup_linear
|
| 35 |
+
|
| 36 |
+
logging.basicConfig(format='%(asctime)s - %(levelname)s - %(name)s - %(message)s',
|
| 37 |
+
datefmt='%m/%d/%Y %H:%M:%S',
|
| 38 |
+
level=logging.INFO)
|
| 39 |
+
logger = logging.getLogger(__name__)
|
| 40 |
+
|
| 41 |
+
|
| 42 |
+
class BERTDataset(Dataset):
|
| 43 |
+
def __init__(self, corpus_path, tokenizer, seq_len, encoding="utf-8", corpus_lines=None, on_memory=True):
|
| 44 |
+
self.vocab = tokenizer.vocab
|
| 45 |
+
self.tokenizer = tokenizer
|
| 46 |
+
self.seq_len = seq_len
|
| 47 |
+
self.on_memory = on_memory
|
| 48 |
+
self.corpus_lines = corpus_lines # number of non-empty lines in input corpus
|
| 49 |
+
self.corpus_path = corpus_path
|
| 50 |
+
self.encoding = encoding
|
| 51 |
+
self.current_doc = 0 # to avoid random sentence from same doc
|
| 52 |
+
|
| 53 |
+
# for loading samples directly from file
|
| 54 |
+
self.sample_counter = 0 # used to keep track of full epochs on file
|
| 55 |
+
self.line_buffer = None # keep second sentence of a pair in memory and use as first sentence in next pair
|
| 56 |
+
|
| 57 |
+
# for loading samples in memory
|
| 58 |
+
self.current_random_doc = 0
|
| 59 |
+
self.num_docs = 0
|
| 60 |
+
self.sample_to_doc = [] # map sample index to doc and line
|
| 61 |
+
|
| 62 |
+
# load samples into memory
|
| 63 |
+
if on_memory:
|
| 64 |
+
self.all_docs = []
|
| 65 |
+
doc = []
|
| 66 |
+
self.corpus_lines = 0
|
| 67 |
+
with open(corpus_path, "r", encoding=encoding) as f:
|
| 68 |
+
for line in tqdm(f, desc="Loading Dataset", total=corpus_lines):
|
| 69 |
+
line = line.strip()
|
| 70 |
+
if line == "":
|
| 71 |
+
self.all_docs.append(doc)
|
| 72 |
+
doc = []
|
| 73 |
+
#remove last added sample because there won't be a subsequent line anymore in the doc
|
| 74 |
+
self.sample_to_doc.pop()
|
| 75 |
+
else:
|
| 76 |
+
#store as one sample
|
| 77 |
+
sample = {"doc_id": len(self.all_docs),
|
| 78 |
+
"line": len(doc)}
|
| 79 |
+
self.sample_to_doc.append(sample)
|
| 80 |
+
doc.append(line)
|
| 81 |
+
self.corpus_lines = self.corpus_lines + 1
|
| 82 |
+
|
| 83 |
+
# if last row in file is not empty
|
| 84 |
+
if self.all_docs[-1] != doc:
|
| 85 |
+
self.all_docs.append(doc)
|
| 86 |
+
self.sample_to_doc.pop()
|
| 87 |
+
|
| 88 |
+
self.num_docs = len(self.all_docs)
|
| 89 |
+
|
| 90 |
+
# load samples later lazily from disk
|
| 91 |
+
else:
|
| 92 |
+
if self.corpus_lines is None:
|
| 93 |
+
with open(corpus_path, "r", encoding=encoding) as f:
|
| 94 |
+
self.corpus_lines = 0
|
| 95 |
+
for line in tqdm(f, desc="Loading Dataset", total=corpus_lines):
|
| 96 |
+
if line.strip() == "":
|
| 97 |
+
self.num_docs += 1
|
| 98 |
+
else:
|
| 99 |
+
self.corpus_lines += 1
|
| 100 |
+
|
| 101 |
+
# if doc does not end with empty line
|
| 102 |
+
if line.strip() != "":
|
| 103 |
+
self.num_docs += 1
|
| 104 |
+
|
| 105 |
+
self.file = open(corpus_path, "r", encoding=encoding)
|
| 106 |
+
self.random_file = open(corpus_path, "r", encoding=encoding)
|
| 107 |
+
|
| 108 |
+
def __len__(self):
|
| 109 |
+
# last line of doc won't be used, because there's no "nextSentence". Additionally, we start counting at 0.
|
| 110 |
+
return self.corpus_lines - self.num_docs - 1
|
| 111 |
+
|
| 112 |
+
def __getitem__(self, item):
|
| 113 |
+
cur_id = self.sample_counter
|
| 114 |
+
self.sample_counter += 1
|
| 115 |
+
if not self.on_memory:
|
| 116 |
+
# after one epoch we start again from beginning of file
|
| 117 |
+
if cur_id != 0 and (cur_id % len(self) == 0):
|
| 118 |
+
self.file.close()
|
| 119 |
+
self.file = open(self.corpus_path, "r", encoding=self.encoding)
|
| 120 |
+
|
| 121 |
+
t1, t2, is_next_label = self.random_sent(item)
|
| 122 |
+
|
| 123 |
+
# tokenize
|
| 124 |
+
tokens_a = self.tokenizer.tokenize(t1)
|
| 125 |
+
tokens_b = self.tokenizer.tokenize(t2)
|
| 126 |
+
|
| 127 |
+
# combine to one sample
|
| 128 |
+
cur_example = InputExample(guid=cur_id, tokens_a=tokens_a, tokens_b=tokens_b, is_next=is_next_label)
|
| 129 |
+
|
| 130 |
+
# transform sample to features
|
| 131 |
+
cur_features = convert_example_to_features(cur_example, self.seq_len, self.tokenizer)
|
| 132 |
+
|
| 133 |
+
cur_tensors = (torch.tensor(cur_features.input_ids),
|
| 134 |
+
torch.tensor(cur_features.input_mask),
|
| 135 |
+
torch.tensor(cur_features.segment_ids),
|
| 136 |
+
torch.tensor(cur_features.lm_label_ids),
|
| 137 |
+
torch.tensor(cur_features.is_next))
|
| 138 |
+
|
| 139 |
+
return cur_tensors
|
| 140 |
+
|
| 141 |
+
def random_sent(self, index):
|
| 142 |
+
"""
|
| 143 |
+
Get one sample from corpus consisting of two sentences. With prob. 50% these are two subsequent sentences
|
| 144 |
+
from one doc. With 50% the second sentence will be a random one from another doc.
|
| 145 |
+
:param index: int, index of sample.
|
| 146 |
+
:return: (str, str, int), sentence 1, sentence 2, isNextSentence Label
|
| 147 |
+
"""
|
| 148 |
+
t1, t2 = self.get_corpus_line(index)
|
| 149 |
+
if random.random() > 0.5:
|
| 150 |
+
label = 0
|
| 151 |
+
else:
|
| 152 |
+
t2 = self.get_random_line()
|
| 153 |
+
label = 1
|
| 154 |
+
|
| 155 |
+
assert len(t1) > 0
|
| 156 |
+
assert len(t2) > 0
|
| 157 |
+
return t1, t2, label
|
| 158 |
+
|
| 159 |
+
def get_corpus_line(self, item):
|
| 160 |
+
"""
|
| 161 |
+
Get one sample from corpus consisting of a pair of two subsequent lines from the same doc.
|
| 162 |
+
:param item: int, index of sample.
|
| 163 |
+
:return: (str, str), two subsequent sentences from corpus
|
| 164 |
+
"""
|
| 165 |
+
t1 = ""
|
| 166 |
+
t2 = ""
|
| 167 |
+
assert item < self.corpus_lines
|
| 168 |
+
if self.on_memory:
|
| 169 |
+
sample = self.sample_to_doc[item]
|
| 170 |
+
t1 = self.all_docs[sample["doc_id"]][sample["line"]]
|
| 171 |
+
t2 = self.all_docs[sample["doc_id"]][sample["line"]+1]
|
| 172 |
+
# used later to avoid random nextSentence from same doc
|
| 173 |
+
self.current_doc = sample["doc_id"]
|
| 174 |
+
return t1, t2
|
| 175 |
+
else:
|
| 176 |
+
if self.line_buffer is None:
|
| 177 |
+
# read first non-empty line of file
|
| 178 |
+
while t1 == "" :
|
| 179 |
+
t1 = next(self.file).strip()
|
| 180 |
+
t2 = next(self.file).strip()
|
| 181 |
+
else:
|
| 182 |
+
# use t2 from previous iteration as new t1
|
| 183 |
+
t1 = self.line_buffer
|
| 184 |
+
t2 = next(self.file).strip()
|
| 185 |
+
# skip empty rows that are used for separating documents and keep track of current doc id
|
| 186 |
+
while t2 == "" or t1 == "":
|
| 187 |
+
t1 = next(self.file).strip()
|
| 188 |
+
t2 = next(self.file).strip()
|
| 189 |
+
self.current_doc = self.current_doc+1
|
| 190 |
+
self.line_buffer = t2
|
| 191 |
+
|
| 192 |
+
assert t1 != ""
|
| 193 |
+
assert t2 != ""
|
| 194 |
+
return t1, t2
|
| 195 |
+
|
| 196 |
+
def get_random_line(self):
|
| 197 |
+
"""
|
| 198 |
+
Get random line from another document for nextSentence task.
|
| 199 |
+
:return: str, content of one line
|
| 200 |
+
"""
|
| 201 |
+
# Similar to original tf repo: This outer loop should rarely go for more than one iteration for large
|
| 202 |
+
# corpora. However, just to be careful, we try to make sure that
|
| 203 |
+
# the random document is not the same as the document we're processing.
|
| 204 |
+
for _ in range(10):
|
| 205 |
+
if self.on_memory:
|
| 206 |
+
rand_doc_idx = random.randint(0, len(self.all_docs)-1)
|
| 207 |
+
rand_doc = self.all_docs[rand_doc_idx]
|
| 208 |
+
line = rand_doc[random.randrange(len(rand_doc))]
|
| 209 |
+
else:
|
| 210 |
+
rand_index = random.randint(1, self.corpus_lines if self.corpus_lines < 1000 else 1000)
|
| 211 |
+
#pick random line
|
| 212 |
+
for _ in range(rand_index):
|
| 213 |
+
line = self.get_next_line()
|
| 214 |
+
#check if our picked random line is really from another doc like we want it to be
|
| 215 |
+
if self.current_random_doc != self.current_doc:
|
| 216 |
+
break
|
| 217 |
+
return line
|
| 218 |
+
|
| 219 |
+
def get_next_line(self):
|
| 220 |
+
""" Gets next line of random_file and starts over when reaching end of file"""
|
| 221 |
+
try:
|
| 222 |
+
line = next(self.random_file).strip()
|
| 223 |
+
#keep track of which document we are currently looking at to later avoid having the same doc as t1
|
| 224 |
+
if line == "":
|
| 225 |
+
self.current_random_doc = self.current_random_doc + 1
|
| 226 |
+
line = next(self.random_file).strip()
|
| 227 |
+
except StopIteration:
|
| 228 |
+
self.random_file.close()
|
| 229 |
+
self.random_file = open(self.corpus_path, "r", encoding=self.encoding)
|
| 230 |
+
line = next(self.random_file).strip()
|
| 231 |
+
return line
|
| 232 |
+
|
| 233 |
+
|
| 234 |
+
class InputExample(object):
|
| 235 |
+
"""A single training/test example for the language model."""
|
| 236 |
+
|
| 237 |
+
def __init__(self, guid, tokens_a, tokens_b=None, is_next=None, lm_labels=None):
|
| 238 |
+
"""Constructs a InputExample.
|
| 239 |
+
|
| 240 |
+
Args:
|
| 241 |
+
guid: Unique id for the example.
|
| 242 |
+
tokens_a: string. The untokenized text of the first sequence. For single
|
| 243 |
+
sequence tasks, only this sequence must be specified.
|
| 244 |
+
tokens_b: (Optional) string. The untokenized text of the second sequence.
|
| 245 |
+
Only must be specified for sequence pair tasks.
|
| 246 |
+
label: (Optional) string. The label of the example. This should be
|
| 247 |
+
specified for train and dev examples, but not for test examples.
|
| 248 |
+
"""
|
| 249 |
+
self.guid = guid
|
| 250 |
+
self.tokens_a = tokens_a
|
| 251 |
+
self.tokens_b = tokens_b
|
| 252 |
+
self.is_next = is_next # nextSentence
|
| 253 |
+
self.lm_labels = lm_labels # masked words for language model
|
| 254 |
+
|
| 255 |
+
|
| 256 |
+
class InputFeatures(object):
|
| 257 |
+
"""A single set of features of data."""
|
| 258 |
+
|
| 259 |
+
def __init__(self, input_ids, input_mask, segment_ids, is_next, lm_label_ids):
|
| 260 |
+
self.input_ids = input_ids
|
| 261 |
+
self.input_mask = input_mask
|
| 262 |
+
self.segment_ids = segment_ids
|
| 263 |
+
self.is_next = is_next
|
| 264 |
+
self.lm_label_ids = lm_label_ids
|
| 265 |
+
|
| 266 |
+
|
| 267 |
+
def random_word(tokens, tokenizer):
|
| 268 |
+
"""
|
| 269 |
+
Masking some random tokens for Language Model task with probabilities as in the original BERT paper.
|
| 270 |
+
:param tokens: list of str, tokenized sentence.
|
| 271 |
+
:param tokenizer: Tokenizer, object used for tokenization (we need it's vocab here)
|
| 272 |
+
:return: (list of str, list of int), masked tokens and related labels for LM prediction
|
| 273 |
+
"""
|
| 274 |
+
output_label = []
|
| 275 |
+
|
| 276 |
+
for i, token in enumerate(tokens):
|
| 277 |
+
prob = random.random()
|
| 278 |
+
# mask token with 15% probability
|
| 279 |
+
if prob < 0.15:
|
| 280 |
+
prob /= 0.15
|
| 281 |
+
|
| 282 |
+
# 80% randomly change token to mask token
|
| 283 |
+
if prob < 0.8:
|
| 284 |
+
tokens[i] = "[MASK]"
|
| 285 |
+
|
| 286 |
+
# 10% randomly change token to random token
|
| 287 |
+
elif prob < 0.9:
|
| 288 |
+
tokens[i] = random.choice(list(tokenizer.vocab.items()))[0]
|
| 289 |
+
|
| 290 |
+
# -> rest 10% randomly keep current token
|
| 291 |
+
|
| 292 |
+
# append current token to output (we will predict these later)
|
| 293 |
+
try:
|
| 294 |
+
output_label.append(tokenizer.vocab[token])
|
| 295 |
+
except KeyError:
|
| 296 |
+
# For unknown words (should not occur with BPE vocab)
|
| 297 |
+
output_label.append(tokenizer.vocab["[UNK]"])
|
| 298 |
+
logger.warning("Cannot find token '{}' in vocab. Using [UNK] insetad".format(token))
|
| 299 |
+
else:
|
| 300 |
+
# no masking token (will be ignored by loss function later)
|
| 301 |
+
output_label.append(-1)
|
| 302 |
+
|
| 303 |
+
return tokens, output_label
|
| 304 |
+
|
| 305 |
+
|
| 306 |
+
def convert_example_to_features(example, max_seq_length, tokenizer):
|
| 307 |
+
"""
|
| 308 |
+
Convert a raw sample (pair of sentences as tokenized strings) into a proper training sample with
|
| 309 |
+
IDs, LM labels, input_mask, CLS and SEP tokens etc.
|
| 310 |
+
:param example: InputExample, containing sentence input as strings and is_next label
|
| 311 |
+
:param max_seq_length: int, maximum length of sequence.
|
| 312 |
+
:param tokenizer: Tokenizer
|
| 313 |
+
:return: InputFeatures, containing all inputs and labels of one sample as IDs (as used for model training)
|
| 314 |
+
"""
|
| 315 |
+
tokens_a = example.tokens_a
|
| 316 |
+
tokens_b = example.tokens_b
|
| 317 |
+
# Modifies `tokens_a` and `tokens_b` in place so that the total
|
| 318 |
+
# length is less than the specified length.
|
| 319 |
+
# Account for [CLS], [SEP], [SEP] with "- 3"
|
| 320 |
+
_truncate_seq_pair(tokens_a, tokens_b, max_seq_length - 3)
|
| 321 |
+
|
| 322 |
+
tokens_a, t1_label = random_word(tokens_a, tokenizer)
|
| 323 |
+
tokens_b, t2_label = random_word(tokens_b, tokenizer)
|
| 324 |
+
# concatenate lm labels and account for CLS, SEP, SEP
|
| 325 |
+
lm_label_ids = ([-1] + t1_label + [-1] + t2_label + [-1])
|
| 326 |
+
|
| 327 |
+
# The convention in BERT is:
|
| 328 |
+
# (a) For sequence pairs:
|
| 329 |
+
# tokens: [CLS] is this jack ##son ##ville ? [SEP] no it is not . [SEP]
|
| 330 |
+
# type_ids: 0 0 0 0 0 0 0 0 1 1 1 1 1 1
|
| 331 |
+
# (b) For single sequences:
|
| 332 |
+
# tokens: [CLS] the dog is hairy . [SEP]
|
| 333 |
+
# type_ids: 0 0 0 0 0 0 0
|
| 334 |
+
#
|
| 335 |
+
# Where "type_ids" are used to indicate whether this is the first
|
| 336 |
+
# sequence or the second sequence. The embedding vectors for `type=0` and
|
| 337 |
+
# `type=1` were learned during pre-training and are added to the wordpiece
|
| 338 |
+
# embedding vector (and position vector). This is not *strictly* necessary
|
| 339 |
+
# since the [SEP] token unambigiously separates the sequences, but it makes
|
| 340 |
+
# it easier for the model to learn the concept of sequences.
|
| 341 |
+
#
|
| 342 |
+
# For classification tasks, the first vector (corresponding to [CLS]) is
|
| 343 |
+
# used as as the "sentence vector". Note that this only makes sense because
|
| 344 |
+
# the entire model is fine-tuned.
|
| 345 |
+
tokens = []
|
| 346 |
+
segment_ids = []
|
| 347 |
+
tokens.append("[CLS]")
|
| 348 |
+
segment_ids.append(0)
|
| 349 |
+
for token in tokens_a:
|
| 350 |
+
tokens.append(token)
|
| 351 |
+
segment_ids.append(0)
|
| 352 |
+
tokens.append("[SEP]")
|
| 353 |
+
segment_ids.append(0)
|
| 354 |
+
|
| 355 |
+
assert len(tokens_b) > 0
|
| 356 |
+
for token in tokens_b:
|
| 357 |
+
tokens.append(token)
|
| 358 |
+
segment_ids.append(1)
|
| 359 |
+
tokens.append("[SEP]")
|
| 360 |
+
segment_ids.append(1)
|
| 361 |
+
|
| 362 |
+
input_ids = tokenizer.convert_tokens_to_ids(tokens)
|
| 363 |
+
|
| 364 |
+
# The mask has 1 for real tokens and 0 for padding tokens. Only real
|
| 365 |
+
# tokens are attended to.
|
| 366 |
+
input_mask = [1] * len(input_ids)
|
| 367 |
+
|
| 368 |
+
# Zero-pad up to the sequence length.
|
| 369 |
+
while len(input_ids) < max_seq_length:
|
| 370 |
+
input_ids.append(0)
|
| 371 |
+
input_mask.append(0)
|
| 372 |
+
segment_ids.append(0)
|
| 373 |
+
lm_label_ids.append(-1)
|
| 374 |
+
|
| 375 |
+
assert len(input_ids) == max_seq_length
|
| 376 |
+
assert len(input_mask) == max_seq_length
|
| 377 |
+
assert len(segment_ids) == max_seq_length
|
| 378 |
+
assert len(lm_label_ids) == max_seq_length
|
| 379 |
+
|
| 380 |
+
if example.guid < 5:
|
| 381 |
+
logger.info("*** Example ***")
|
| 382 |
+
logger.info("guid: %s" % (example.guid))
|
| 383 |
+
logger.info("tokens: %s" % " ".join(
|
| 384 |
+
[str(x) for x in tokens]))
|
| 385 |
+
logger.info("input_ids: %s" % " ".join([str(x) for x in input_ids]))
|
| 386 |
+
logger.info("input_mask: %s" % " ".join([str(x) for x in input_mask]))
|
| 387 |
+
logger.info(
|
| 388 |
+
"segment_ids: %s" % " ".join([str(x) for x in segment_ids]))
|
| 389 |
+
logger.info("LM label: %s " % (lm_label_ids))
|
| 390 |
+
logger.info("Is next sentence label: %s " % (example.is_next))
|
| 391 |
+
|
| 392 |
+
features = InputFeatures(input_ids=input_ids,
|
| 393 |
+
input_mask=input_mask,
|
| 394 |
+
segment_ids=segment_ids,
|
| 395 |
+
lm_label_ids=lm_label_ids,
|
| 396 |
+
is_next=example.is_next)
|
| 397 |
+
return features
|
| 398 |
+
|
| 399 |
+
|
| 400 |
+
def main():
|
| 401 |
+
parser = argparse.ArgumentParser()
|
| 402 |
+
|
| 403 |
+
## Required parameters
|
| 404 |
+
parser.add_argument("--train_corpus",
|
| 405 |
+
default=None,
|
| 406 |
+
type=str,
|
| 407 |
+
required=True,
|
| 408 |
+
help="The input train corpus.")
|
| 409 |
+
parser.add_argument("--bert_model", default=None, type=str, required=True,
|
| 410 |
+
help="Bert pre-trained model selected in the list: bert-base-uncased, "
|
| 411 |
+
"bert-large-uncased, bert-base-cased, bert-base-multilingual, bert-base-chinese.")
|
| 412 |
+
parser.add_argument("--output_dir",
|
| 413 |
+
default=None,
|
| 414 |
+
type=str,
|
| 415 |
+
required=True,
|
| 416 |
+
help="The output directory where the model checkpoints will be written.")
|
| 417 |
+
|
| 418 |
+
## Other parameters
|
| 419 |
+
parser.add_argument("--max_seq_length",
|
| 420 |
+
default=128,
|
| 421 |
+
type=int,
|
| 422 |
+
help="The maximum total input sequence length after WordPiece tokenization. \n"
|
| 423 |
+
"Sequences longer than this will be truncated, and sequences shorter \n"
|
| 424 |
+
"than this will be padded.")
|
| 425 |
+
parser.add_argument("--do_train",
|
| 426 |
+
action='store_true',
|
| 427 |
+
help="Whether to run training.")
|
| 428 |
+
parser.add_argument("--train_batch_size",
|
| 429 |
+
default=32,
|
| 430 |
+
type=int,
|
| 431 |
+
help="Total batch size for training.")
|
| 432 |
+
parser.add_argument("--learning_rate",
|
| 433 |
+
default=3e-5,
|
| 434 |
+
type=float,
|
| 435 |
+
help="The initial learning rate for Adam.")
|
| 436 |
+
parser.add_argument("--num_train_epochs",
|
| 437 |
+
default=3.0,
|
| 438 |
+
type=float,
|
| 439 |
+
help="Total number of training epochs to perform.")
|
| 440 |
+
parser.add_argument("--warmup_proportion",
|
| 441 |
+
default=0.1,
|
| 442 |
+
type=float,
|
| 443 |
+
help="Proportion of training to perform linear learning rate warmup for. "
|
| 444 |
+
"E.g., 0.1 = 10%% of training.")
|
| 445 |
+
parser.add_argument("--no_cuda",
|
| 446 |
+
action='store_true',
|
| 447 |
+
help="Whether not to use CUDA when available")
|
| 448 |
+
parser.add_argument("--on_memory",
|
| 449 |
+
action='store_true',
|
| 450 |
+
help="Whether to load train samples into memory or use disk")
|
| 451 |
+
parser.add_argument("--do_lower_case",
|
| 452 |
+
action='store_true',
|
| 453 |
+
help="Whether to lower case the input text. True for uncased models, False for cased models.")
|
| 454 |
+
parser.add_argument("--local_rank",
|
| 455 |
+
type=int,
|
| 456 |
+
default=-1,
|
| 457 |
+
help="local_rank for distributed training on gpus")
|
| 458 |
+
parser.add_argument('--seed',
|
| 459 |
+
type=int,
|
| 460 |
+
default=42,
|
| 461 |
+
help="random seed for initialization")
|
| 462 |
+
parser.add_argument('--gradient_accumulation_steps',
|
| 463 |
+
type=int,
|
| 464 |
+
default=1,
|
| 465 |
+
help="Number of updates steps to accumualte before performing a backward/update pass.")
|
| 466 |
+
parser.add_argument('--fp16',
|
| 467 |
+
action='store_true',
|
| 468 |
+
help="Whether to use 16-bit float precision instead of 32-bit")
|
| 469 |
+
parser.add_argument('--loss_scale',
|
| 470 |
+
type = float, default = 0,
|
| 471 |
+
help = "Loss scaling to improve fp16 numeric stability. Only used when fp16 set to True.\n"
|
| 472 |
+
"0 (default value): dynamic loss scaling.\n"
|
| 473 |
+
"Positive power of 2: static loss scaling value.\n")
|
| 474 |
+
|
| 475 |
+
args = parser.parse_args()
|
| 476 |
+
|
| 477 |
+
if args.local_rank == -1 or args.no_cuda:
|
| 478 |
+
device = torch.device("cuda" if torch.cuda.is_available() and not args.no_cuda else "cpu")
|
| 479 |
+
n_gpu = torch.cuda.device_count()
|
| 480 |
+
else:
|
| 481 |
+
torch.cuda.set_device(args.local_rank)
|
| 482 |
+
device = torch.device("cuda", args.local_rank)
|
| 483 |
+
n_gpu = 1
|
| 484 |
+
# Initializes the distributed backend which will take care of sychronizing nodes/GPUs
|
| 485 |
+
torch.distributed.init_process_group(backend='nccl')
|
| 486 |
+
logger.info("device: {} n_gpu: {}, distributed training: {}, 16-bits training: {}".format(
|
| 487 |
+
device, n_gpu, bool(args.local_rank != -1), args.fp16))
|
| 488 |
+
|
| 489 |
+
if args.gradient_accumulation_steps < 1:
|
| 490 |
+
raise ValueError("Invalid gradient_accumulation_steps parameter: {}, should be >= 1".format(
|
| 491 |
+
args.gradient_accumulation_steps))
|
| 492 |
+
|
| 493 |
+
args.train_batch_size = args.train_batch_size // args.gradient_accumulation_steps
|
| 494 |
+
|
| 495 |
+
random.seed(args.seed)
|
| 496 |
+
np.random.seed(args.seed)
|
| 497 |
+
torch.manual_seed(args.seed)
|
| 498 |
+
if n_gpu > 0:
|
| 499 |
+
torch.cuda.manual_seed_all(args.seed)
|
| 500 |
+
|
| 501 |
+
if not args.do_train:
|
| 502 |
+
raise ValueError("Training is currently the only implemented execution option. Please set `do_train`.")
|
| 503 |
+
|
| 504 |
+
if os.path.exists(args.output_dir) and os.listdir(args.output_dir):
|
| 505 |
+
raise ValueError("Output directory ({}) already exists and is not empty.".format(args.output_dir))
|
| 506 |
+
if not os.path.exists(args.output_dir):
|
| 507 |
+
os.makedirs(args.output_dir)
|
| 508 |
+
|
| 509 |
+
tokenizer = BertTokenizer.from_pretrained(args.bert_model, do_lower_case=args.do_lower_case)
|
| 510 |
+
|
| 511 |
+
#train_examples = None
|
| 512 |
+
num_train_optimization_steps = None
|
| 513 |
+
if args.do_train:
|
| 514 |
+
print("Loading Train Dataset", args.train_corpus)
|
| 515 |
+
train_dataset = BERTDataset(args.train_corpus, tokenizer, seq_len=args.max_seq_length,
|
| 516 |
+
corpus_lines=None, on_memory=args.on_memory)
|
| 517 |
+
num_train_optimization_steps = int(
|
| 518 |
+
len(train_dataset) / args.train_batch_size / args.gradient_accumulation_steps) * args.num_train_epochs
|
| 519 |
+
if args.local_rank != -1:
|
| 520 |
+
num_train_optimization_steps = num_train_optimization_steps // torch.distributed.get_world_size()
|
| 521 |
+
|
| 522 |
+
# Prepare model
|
| 523 |
+
model = BertForPreTraining.from_pretrained(args.bert_model)
|
| 524 |
+
if args.fp16:
|
| 525 |
+
model.half()
|
| 526 |
+
model.to(device)
|
| 527 |
+
if args.local_rank != -1:
|
| 528 |
+
try:
|
| 529 |
+
from apex.parallel import DistributedDataParallel as DDP
|
| 530 |
+
except ImportError:
|
| 531 |
+
raise ImportError("Please install apex from https://www.github.com/nvidia/apex to use distributed and fp16 training.")
|
| 532 |
+
model = DDP(model)
|
| 533 |
+
elif n_gpu > 1:
|
| 534 |
+
model = torch.nn.DataParallel(model)
|
| 535 |
+
|
| 536 |
+
# Prepare optimizer
|
| 537 |
+
param_optimizer = list(model.named_parameters())
|
| 538 |
+
no_decay = ['bias', 'LayerNorm.bias', 'LayerNorm.weight']
|
| 539 |
+
optimizer_grouped_parameters = [
|
| 540 |
+
{'params': [p for n, p in param_optimizer if not any(nd in n for nd in no_decay)], 'weight_decay': 0.01},
|
| 541 |
+
{'params': [p for n, p in param_optimizer if any(nd in n for nd in no_decay)], 'weight_decay': 0.0}
|
| 542 |
+
]
|
| 543 |
+
|
| 544 |
+
if args.fp16:
|
| 545 |
+
try:
|
| 546 |
+
from apex.optimizers import FP16_Optimizer
|
| 547 |
+
from apex.optimizers import FusedAdam
|
| 548 |
+
except ImportError:
|
| 549 |
+
raise ImportError("Please install apex from https://www.github.com/nvidia/apex to use distributed and fp16 training.")
|
| 550 |
+
|
| 551 |
+
optimizer = FusedAdam(optimizer_grouped_parameters,
|
| 552 |
+
lr=args.learning_rate,
|
| 553 |
+
bias_correction=False,
|
| 554 |
+
max_grad_norm=1.0)
|
| 555 |
+
if args.loss_scale == 0:
|
| 556 |
+
optimizer = FP16_Optimizer(optimizer, dynamic_loss_scale=True)
|
| 557 |
+
else:
|
| 558 |
+
optimizer = FP16_Optimizer(optimizer, static_loss_scale=args.loss_scale)
|
| 559 |
+
|
| 560 |
+
else:
|
| 561 |
+
optimizer = BertAdam(optimizer_grouped_parameters,
|
| 562 |
+
lr=args.learning_rate,
|
| 563 |
+
warmup=args.warmup_proportion,
|
| 564 |
+
t_total=num_train_optimization_steps)
|
| 565 |
+
|
| 566 |
+
global_step = 0
|
| 567 |
+
if args.do_train:
|
| 568 |
+
logger.info("***** Running training *****")
|
| 569 |
+
logger.info(" Num examples = %d", len(train_dataset))
|
| 570 |
+
logger.info(" Batch size = %d", args.train_batch_size)
|
| 571 |
+
logger.info(" Num steps = %d", num_train_optimization_steps)
|
| 572 |
+
|
| 573 |
+
if args.local_rank == -1:
|
| 574 |
+
train_sampler = RandomSampler(train_dataset)
|
| 575 |
+
else:
|
| 576 |
+
#TODO: check if this works with current data generator from disk that relies on next(file)
|
| 577 |
+
# (it doesn't return item back by index)
|
| 578 |
+
train_sampler = DistributedSampler(train_dataset)
|
| 579 |
+
train_dataloader = DataLoader(train_dataset, sampler=train_sampler, batch_size=args.train_batch_size)
|
| 580 |
+
|
| 581 |
+
model.train()
|
| 582 |
+
for _ in trange(int(args.num_train_epochs), desc="Epoch"):
|
| 583 |
+
tr_loss = 0
|
| 584 |
+
nb_tr_examples, nb_tr_steps = 0, 0
|
| 585 |
+
for step, batch in enumerate(tqdm(train_dataloader, desc="Iteration")):
|
| 586 |
+
batch = tuple(t.to(device) for t in batch)
|
| 587 |
+
input_ids, input_mask, segment_ids, lm_label_ids, is_next = batch
|
| 588 |
+
loss = model(input_ids, segment_ids, input_mask, lm_label_ids, is_next)
|
| 589 |
+
if n_gpu > 1:
|
| 590 |
+
loss = loss.mean() # mean() to average on multi-gpu.
|
| 591 |
+
if args.gradient_accumulation_steps > 1:
|
| 592 |
+
loss = loss / args.gradient_accumulation_steps
|
| 593 |
+
if args.fp16:
|
| 594 |
+
optimizer.backward(loss)
|
| 595 |
+
else:
|
| 596 |
+
loss.backward()
|
| 597 |
+
tr_loss += loss.item()
|
| 598 |
+
nb_tr_examples += input_ids.size(0)
|
| 599 |
+
nb_tr_steps += 1
|
| 600 |
+
if (step + 1) % args.gradient_accumulation_steps == 0:
|
| 601 |
+
if args.fp16:
|
| 602 |
+
# modify learning rate with special warm up BERT uses
|
| 603 |
+
# if args.fp16 is False, BertAdam is used that handles this automatically
|
| 604 |
+
lr_this_step = args.learning_rate * warmup_linear(global_step/num_train_optimization_steps, args.warmup_proportion)
|
| 605 |
+
for param_group in optimizer.param_groups:
|
| 606 |
+
param_group['lr'] = lr_this_step
|
| 607 |
+
optimizer.step()
|
| 608 |
+
optimizer.zero_grad()
|
| 609 |
+
global_step += 1
|
| 610 |
+
|
| 611 |
+
# Save a trained model
|
| 612 |
+
logger.info("** ** * Saving fine - tuned model ** ** * ")
|
| 613 |
+
model_to_save = model.module if hasattr(model, 'module') else model # Only save the model it-self
|
| 614 |
+
output_model_file = os.path.join(args.output_dir, "pytorch_model.bin")
|
| 615 |
+
if args.do_train:
|
| 616 |
+
torch.save(model_to_save.state_dict(), output_model_file)
|
| 617 |
+
|
| 618 |
+
|
| 619 |
+
def _truncate_seq_pair(tokens_a, tokens_b, max_length):
|
| 620 |
+
"""Truncates a sequence pair in place to the maximum length."""
|
| 621 |
+
|
| 622 |
+
# This is a simple heuristic which will always truncate the longer sequence
|
| 623 |
+
# one token at a time. This makes more sense than truncating an equal percent
|
| 624 |
+
# of tokens from each, since if one sequence is very short then each token
|
| 625 |
+
# that's truncated likely contains more information than a longer sequence.
|
| 626 |
+
while True:
|
| 627 |
+
total_length = len(tokens_a) + len(tokens_b)
|
| 628 |
+
if total_length <= max_length:
|
| 629 |
+
break
|
| 630 |
+
if len(tokens_a) > len(tokens_b):
|
| 631 |
+
tokens_a.pop()
|
| 632 |
+
else:
|
| 633 |
+
tokens_b.pop()
|
| 634 |
+
|
| 635 |
+
|
| 636 |
+
def accuracy(out, labels):
|
| 637 |
+
outputs = np.argmax(out, axis=1)
|
| 638 |
+
return np.sum(outputs == labels)
|
| 639 |
+
|
| 640 |
+
|
| 641 |
+
if __name__ == "__main__":
|
| 642 |
+
main()
|
dark-secrets-of-BERT-master/dark-secrets-of-BERT-master/examples/run_classifier.py
ADDED
|
@@ -0,0 +1,1047 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# coding=utf-8
|
| 2 |
+
# Copyright 2018 The Google AI Language Team Authors and The HuggingFace Inc. team.
|
| 3 |
+
# Copyright (c) 2018, NVIDIA CORPORATION. All rights reserved.
|
| 4 |
+
#
|
| 5 |
+
# Licensed under the Apache License, Version 2.0 (the "License");
|
| 6 |
+
# you may not use this file except in compliance with the License.
|
| 7 |
+
# You may obtain a copy of the License at
|
| 8 |
+
#
|
| 9 |
+
# http://www.apache.org/licenses/LICENSE-2.0
|
| 10 |
+
#
|
| 11 |
+
# Unless required by applicable law or agreed to in writing, software
|
| 12 |
+
# distributed under the License is distributed on an "AS IS" BASIS,
|
| 13 |
+
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
| 14 |
+
# See the License for the specific language governing permissions and
|
| 15 |
+
# limitations under the License.
|
| 16 |
+
"""BERT finetuning runner."""
|
| 17 |
+
|
| 18 |
+
from __future__ import absolute_import, division, print_function
|
| 19 |
+
|
| 20 |
+
import sys
|
| 21 |
+
sys.path.insert(0, "/home/okovaleva/projects/bert_attention/pretrained_bert/pytorch-pretrained-BERT")
|
| 22 |
+
print(sys.path)
|
| 23 |
+
|
| 24 |
+
|
| 25 |
+
import argparse
|
| 26 |
+
import csv
|
| 27 |
+
import logging
|
| 28 |
+
import os
|
| 29 |
+
import random
|
| 30 |
+
import sys
|
| 31 |
+
|
| 32 |
+
import numpy as np
|
| 33 |
+
import torch
|
| 34 |
+
from torch.utils.data import (DataLoader, RandomSampler, SequentialSampler,
|
| 35 |
+
TensorDataset)
|
| 36 |
+
from torch.utils.data.distributed import DistributedSampler
|
| 37 |
+
from tqdm import tqdm, trange
|
| 38 |
+
|
| 39 |
+
from torch.nn import CrossEntropyLoss, MSELoss
|
| 40 |
+
from scipy.stats import pearsonr, spearmanr
|
| 41 |
+
from sklearn.metrics import matthews_corrcoef, f1_score
|
| 42 |
+
|
| 43 |
+
from pytorch_pretrained_bert.file_utils import PYTORCH_PRETRAINED_BERT_CACHE, WEIGHTS_NAME, CONFIG_NAME
|
| 44 |
+
from pytorch_pretrained_bert.modeling import BertForSequenceClassification, BertConfig
|
| 45 |
+
from pytorch_pretrained_bert.tokenization import BertTokenizer
|
| 46 |
+
from pytorch_pretrained_bert.optimization import BertAdam, warmup_linear
|
| 47 |
+
|
| 48 |
+
logger = logging.getLogger(__name__)
|
| 49 |
+
|
| 50 |
+
|
| 51 |
+
class InputExample(object):
|
| 52 |
+
"""A single training/test example for simple sequence classification."""
|
| 53 |
+
|
| 54 |
+
def __init__(self, guid, text_a, text_b=None, label=None):
|
| 55 |
+
"""Constructs a InputExample.
|
| 56 |
+
|
| 57 |
+
Args:
|
| 58 |
+
guid: Unique id for the example.
|
| 59 |
+
text_a: string. The untokenized text of the first sequence. For single
|
| 60 |
+
sequence tasks, only this sequence must be specified.
|
| 61 |
+
text_b: (Optional) string. The untokenized text of the second sequence.
|
| 62 |
+
Only must be specified for sequence pair tasks.
|
| 63 |
+
label: (Optional) string. The label of the example. This should be
|
| 64 |
+
specified for train and dev examples, but not for test examples.
|
| 65 |
+
"""
|
| 66 |
+
self.guid = guid
|
| 67 |
+
self.text_a = text_a
|
| 68 |
+
self.text_b = text_b
|
| 69 |
+
self.label = label
|
| 70 |
+
|
| 71 |
+
|
| 72 |
+
class InputFeatures(object):
|
| 73 |
+
"""A single set of features of data."""
|
| 74 |
+
|
| 75 |
+
def __init__(self, input_ids, input_mask, segment_ids, label_id):
|
| 76 |
+
self.input_ids = input_ids
|
| 77 |
+
self.input_mask = input_mask
|
| 78 |
+
self.segment_ids = segment_ids
|
| 79 |
+
self.label_id = label_id
|
| 80 |
+
|
| 81 |
+
|
| 82 |
+
class DataProcessor(object):
|
| 83 |
+
"""Base class for data converters for sequence classification data sets."""
|
| 84 |
+
|
| 85 |
+
def get_train_examples(self, data_dir):
|
| 86 |
+
"""Gets a collection of `InputExample`s for the train set."""
|
| 87 |
+
raise NotImplementedError()
|
| 88 |
+
|
| 89 |
+
def get_dev_examples(self, data_dir):
|
| 90 |
+
"""Gets a collection of `InputExample`s for the dev set."""
|
| 91 |
+
raise NotImplementedError()
|
| 92 |
+
|
| 93 |
+
def get_labels(self):
|
| 94 |
+
"""Gets the list of labels for this data set."""
|
| 95 |
+
raise NotImplementedError()
|
| 96 |
+
|
| 97 |
+
@classmethod
|
| 98 |
+
def _read_tsv(cls, input_file, quotechar=None):
|
| 99 |
+
"""Reads a tab separated value file."""
|
| 100 |
+
with open(input_file, "r", encoding="utf-8") as f:
|
| 101 |
+
reader = csv.reader(f, delimiter="\t", quotechar=quotechar)
|
| 102 |
+
lines = []
|
| 103 |
+
for line in reader:
|
| 104 |
+
if sys.version_info[0] == 2:
|
| 105 |
+
line = list(unicode(cell, 'utf-8') for cell in line)
|
| 106 |
+
lines.append(line)
|
| 107 |
+
return lines
|
| 108 |
+
|
| 109 |
+
|
| 110 |
+
class MrpcProcessor(DataProcessor):
|
| 111 |
+
"""Processor for the MRPC data set (GLUE version)."""
|
| 112 |
+
|
| 113 |
+
def get_train_examples(self, data_dir):
|
| 114 |
+
"""See base class."""
|
| 115 |
+
logger.info("LOOKING AT {}".format(os.path.join(data_dir, "train.tsv")))
|
| 116 |
+
return self._create_examples(
|
| 117 |
+
self._read_tsv(os.path.join(data_dir, "train.tsv")), "train")
|
| 118 |
+
|
| 119 |
+
def get_dev_examples(self, data_dir):
|
| 120 |
+
"""See base class."""
|
| 121 |
+
return self._create_examples(
|
| 122 |
+
self._read_tsv(os.path.join(data_dir, "dev.tsv")), "dev")
|
| 123 |
+
|
| 124 |
+
def get_labels(self):
|
| 125 |
+
"""See base class."""
|
| 126 |
+
return ["0", "1"]
|
| 127 |
+
|
| 128 |
+
def _create_examples(self, lines, set_type):
|
| 129 |
+
"""Creates examples for the training and dev sets."""
|
| 130 |
+
examples = []
|
| 131 |
+
for (i, line) in enumerate(lines):
|
| 132 |
+
if i == 0:
|
| 133 |
+
continue
|
| 134 |
+
guid = "%s-%s" % (set_type, i)
|
| 135 |
+
text_a = line[3]
|
| 136 |
+
text_b = line[4]
|
| 137 |
+
label = line[0]
|
| 138 |
+
examples.append(
|
| 139 |
+
InputExample(guid=guid, text_a=text_a, text_b=text_b, label=label))
|
| 140 |
+
return examples
|
| 141 |
+
|
| 142 |
+
|
| 143 |
+
class MnliProcessor(DataProcessor):
|
| 144 |
+
"""Processor for the MultiNLI data set (GLUE version)."""
|
| 145 |
+
|
| 146 |
+
def get_train_examples(self, data_dir):
|
| 147 |
+
"""See base class."""
|
| 148 |
+
return self._create_examples(
|
| 149 |
+
self._read_tsv(os.path.join(data_dir, "train.tsv")), "train")
|
| 150 |
+
|
| 151 |
+
def get_dev_examples(self, data_dir):
|
| 152 |
+
"""See base class."""
|
| 153 |
+
return self._create_examples(
|
| 154 |
+
self._read_tsv(os.path.join(data_dir, "dev_matched.tsv")),
|
| 155 |
+
"dev_matched")
|
| 156 |
+
|
| 157 |
+
def get_labels(self):
|
| 158 |
+
"""See base class."""
|
| 159 |
+
return ["contradiction", "entailment", "neutral"]
|
| 160 |
+
|
| 161 |
+
def _create_examples(self, lines, set_type):
|
| 162 |
+
"""Creates examples for the training and dev sets."""
|
| 163 |
+
examples = []
|
| 164 |
+
for (i, line) in enumerate(lines):
|
| 165 |
+
if i == 0:
|
| 166 |
+
continue
|
| 167 |
+
guid = "%s-%s" % (set_type, line[0])
|
| 168 |
+
text_a = line[8]
|
| 169 |
+
text_b = line[9]
|
| 170 |
+
label = line[-1]
|
| 171 |
+
examples.append(
|
| 172 |
+
InputExample(guid=guid, text_a=text_a, text_b=text_b, label=label))
|
| 173 |
+
return examples
|
| 174 |
+
|
| 175 |
+
|
| 176 |
+
class MnliMismatchedProcessor(MnliProcessor):
|
| 177 |
+
"""Processor for the MultiNLI Mismatched data set (GLUE version)."""
|
| 178 |
+
|
| 179 |
+
def get_dev_examples(self, data_dir):
|
| 180 |
+
"""See base class."""
|
| 181 |
+
return self._create_examples(
|
| 182 |
+
self._read_tsv(os.path.join(data_dir, "dev_mismatched.tsv")),
|
| 183 |
+
"dev_matched")
|
| 184 |
+
|
| 185 |
+
|
| 186 |
+
class ColaProcessor(DataProcessor):
|
| 187 |
+
"""Processor for the CoLA data set (GLUE version)."""
|
| 188 |
+
|
| 189 |
+
def get_train_examples(self, data_dir):
|
| 190 |
+
"""See base class."""
|
| 191 |
+
return self._create_examples(
|
| 192 |
+
self._read_tsv(os.path.join(data_dir, "train.tsv")), "train")
|
| 193 |
+
|
| 194 |
+
def get_dev_examples(self, data_dir):
|
| 195 |
+
"""See base class."""
|
| 196 |
+
return self._create_examples(
|
| 197 |
+
self._read_tsv(os.path.join(data_dir, "dev.tsv")), "dev")
|
| 198 |
+
|
| 199 |
+
def get_labels(self):
|
| 200 |
+
"""See base class."""
|
| 201 |
+
return ["0", "1"]
|
| 202 |
+
|
| 203 |
+
def _create_examples(self, lines, set_type):
|
| 204 |
+
"""Creates examples for the training and dev sets."""
|
| 205 |
+
examples = []
|
| 206 |
+
for (i, line) in enumerate(lines):
|
| 207 |
+
guid = "%s-%s" % (set_type, i)
|
| 208 |
+
text_a = line[3]
|
| 209 |
+
label = line[1]
|
| 210 |
+
examples.append(
|
| 211 |
+
InputExample(guid=guid, text_a=text_a, text_b=None, label=label))
|
| 212 |
+
return examples
|
| 213 |
+
|
| 214 |
+
|
| 215 |
+
class Sst2Processor(DataProcessor):
|
| 216 |
+
"""Processor for the SST-2 data set (GLUE version)."""
|
| 217 |
+
|
| 218 |
+
def get_train_examples(self, data_dir):
|
| 219 |
+
"""See base class."""
|
| 220 |
+
return self._create_examples(
|
| 221 |
+
self._read_tsv(os.path.join(data_dir, "train.tsv")), "train")
|
| 222 |
+
|
| 223 |
+
def get_dev_examples(self, data_dir):
|
| 224 |
+
"""See base class."""
|
| 225 |
+
return self._create_examples(
|
| 226 |
+
self._read_tsv(os.path.join(data_dir, "dev.tsv")), "dev")
|
| 227 |
+
|
| 228 |
+
def get_labels(self):
|
| 229 |
+
"""See base class."""
|
| 230 |
+
return ["0", "1"]
|
| 231 |
+
|
| 232 |
+
def _create_examples(self, lines, set_type):
|
| 233 |
+
"""Creates examples for the training and dev sets."""
|
| 234 |
+
examples = []
|
| 235 |
+
for (i, line) in enumerate(lines):
|
| 236 |
+
if i == 0:
|
| 237 |
+
continue
|
| 238 |
+
guid = "%s-%s" % (set_type, i)
|
| 239 |
+
text_a = line[0]
|
| 240 |
+
label = line[1]
|
| 241 |
+
examples.append(
|
| 242 |
+
InputExample(guid=guid, text_a=text_a, text_b=None, label=label))
|
| 243 |
+
return examples
|
| 244 |
+
|
| 245 |
+
|
| 246 |
+
class StsbProcessor(DataProcessor):
|
| 247 |
+
"""Processor for the STS-B data set (GLUE version)."""
|
| 248 |
+
|
| 249 |
+
def get_train_examples(self, data_dir):
|
| 250 |
+
"""See base class."""
|
| 251 |
+
return self._create_examples(
|
| 252 |
+
self._read_tsv(os.path.join(data_dir, "train.tsv")), "train")
|
| 253 |
+
|
| 254 |
+
def get_dev_examples(self, data_dir):
|
| 255 |
+
"""See base class."""
|
| 256 |
+
return self._create_examples(
|
| 257 |
+
self._read_tsv(os.path.join(data_dir, "dev.tsv")), "dev")
|
| 258 |
+
|
| 259 |
+
def get_labels(self):
|
| 260 |
+
"""See base class."""
|
| 261 |
+
return [None]
|
| 262 |
+
|
| 263 |
+
def _create_examples(self, lines, set_type):
|
| 264 |
+
"""Creates examples for the training and dev sets."""
|
| 265 |
+
examples = []
|
| 266 |
+
for (i, line) in enumerate(lines):
|
| 267 |
+
if i == 0:
|
| 268 |
+
continue
|
| 269 |
+
guid = "%s-%s" % (set_type, line[0])
|
| 270 |
+
text_a = line[7]
|
| 271 |
+
text_b = line[8]
|
| 272 |
+
label = line[-1]
|
| 273 |
+
examples.append(
|
| 274 |
+
InputExample(guid=guid, text_a=text_a, text_b=text_b, label=label))
|
| 275 |
+
return examples
|
| 276 |
+
|
| 277 |
+
|
| 278 |
+
class QqpProcessor(DataProcessor):
|
| 279 |
+
"""Processor for the STS-B data set (GLUE version)."""
|
| 280 |
+
|
| 281 |
+
def get_train_examples(self, data_dir):
|
| 282 |
+
"""See base class."""
|
| 283 |
+
return self._create_examples(
|
| 284 |
+
self._read_tsv(os.path.join(data_dir, "train.tsv")), "train")
|
| 285 |
+
|
| 286 |
+
def get_dev_examples(self, data_dir):
|
| 287 |
+
"""See base class."""
|
| 288 |
+
return self._create_examples(
|
| 289 |
+
self._read_tsv(os.path.join(data_dir, "dev.tsv")), "dev")
|
| 290 |
+
|
| 291 |
+
def get_labels(self):
|
| 292 |
+
"""See base class."""
|
| 293 |
+
return ["0", "1"]
|
| 294 |
+
|
| 295 |
+
def _create_examples(self, lines, set_type):
|
| 296 |
+
"""Creates examples for the training and dev sets."""
|
| 297 |
+
examples = []
|
| 298 |
+
for (i, line) in enumerate(lines):
|
| 299 |
+
if i == 0:
|
| 300 |
+
continue
|
| 301 |
+
guid = "%s-%s" % (set_type, line[0])
|
| 302 |
+
try:
|
| 303 |
+
text_a = line[3]
|
| 304 |
+
text_b = line[4]
|
| 305 |
+
label = line[5]
|
| 306 |
+
except IndexError:
|
| 307 |
+
continue
|
| 308 |
+
examples.append(
|
| 309 |
+
InputExample(guid=guid, text_a=text_a, text_b=text_b, label=label))
|
| 310 |
+
return examples
|
| 311 |
+
|
| 312 |
+
|
| 313 |
+
class QnliProcessor(DataProcessor):
|
| 314 |
+
"""Processor for the STS-B data set (GLUE version)."""
|
| 315 |
+
|
| 316 |
+
def get_train_examples(self, data_dir):
|
| 317 |
+
"""See base class."""
|
| 318 |
+
return self._create_examples(
|
| 319 |
+
self._read_tsv(os.path.join(data_dir, "train.tsv")), "train")
|
| 320 |
+
|
| 321 |
+
def get_dev_examples(self, data_dir):
|
| 322 |
+
"""See base class."""
|
| 323 |
+
return self._create_examples(
|
| 324 |
+
self._read_tsv(os.path.join(data_dir, "dev.tsv")),
|
| 325 |
+
"dev_matched")
|
| 326 |
+
|
| 327 |
+
def get_labels(self):
|
| 328 |
+
"""See base class."""
|
| 329 |
+
return ["entailment", "not_entailment"]
|
| 330 |
+
|
| 331 |
+
def _create_examples(self, lines, set_type):
|
| 332 |
+
"""Creates examples for the training and dev sets."""
|
| 333 |
+
examples = []
|
| 334 |
+
for (i, line) in enumerate(lines):
|
| 335 |
+
if i == 0:
|
| 336 |
+
continue
|
| 337 |
+
guid = "%s-%s" % (set_type, line[0])
|
| 338 |
+
text_a = line[1]
|
| 339 |
+
text_b = line[2]
|
| 340 |
+
label = line[-1]
|
| 341 |
+
examples.append(
|
| 342 |
+
InputExample(guid=guid, text_a=text_a, text_b=text_b, label=label))
|
| 343 |
+
return examples
|
| 344 |
+
|
| 345 |
+
|
| 346 |
+
class RteProcessor(DataProcessor):
|
| 347 |
+
"""Processor for the RTE data set (GLUE version)."""
|
| 348 |
+
|
| 349 |
+
def get_train_examples(self, data_dir):
|
| 350 |
+
"""See base class."""
|
| 351 |
+
return self._create_examples(
|
| 352 |
+
self._read_tsv(os.path.join(data_dir, "train.tsv")), "train")
|
| 353 |
+
|
| 354 |
+
def get_dev_examples(self, data_dir):
|
| 355 |
+
"""See base class."""
|
| 356 |
+
return self._create_examples(
|
| 357 |
+
self._read_tsv(os.path.join(data_dir, "dev.tsv")), "dev")
|
| 358 |
+
|
| 359 |
+
def get_labels(self):
|
| 360 |
+
"""See base class."""
|
| 361 |
+
return ["entailment", "not_entailment"]
|
| 362 |
+
|
| 363 |
+
def _create_examples(self, lines, set_type):
|
| 364 |
+
"""Creates examples for the training and dev sets."""
|
| 365 |
+
examples = []
|
| 366 |
+
for (i, line) in enumerate(lines):
|
| 367 |
+
if i == 0:
|
| 368 |
+
continue
|
| 369 |
+
guid = "%s-%s" % (set_type, line[0])
|
| 370 |
+
text_a = line[1]
|
| 371 |
+
text_b = line[2]
|
| 372 |
+
label = line[-1]
|
| 373 |
+
examples.append(
|
| 374 |
+
InputExample(guid=guid, text_a=text_a, text_b=text_b, label=label))
|
| 375 |
+
return examples
|
| 376 |
+
|
| 377 |
+
|
| 378 |
+
class WnliProcessor(DataProcessor):
|
| 379 |
+
"""Processor for the WNLI data set (GLUE version)."""
|
| 380 |
+
|
| 381 |
+
def get_train_examples(self, data_dir):
|
| 382 |
+
"""See base class."""
|
| 383 |
+
return self._create_examples(
|
| 384 |
+
self._read_tsv(os.path.join(data_dir, "train.tsv")), "train")
|
| 385 |
+
|
| 386 |
+
def get_dev_examples(self, data_dir):
|
| 387 |
+
"""See base class."""
|
| 388 |
+
return self._create_examples(
|
| 389 |
+
self._read_tsv(os.path.join(data_dir, "dev.tsv")), "dev")
|
| 390 |
+
|
| 391 |
+
def get_labels(self):
|
| 392 |
+
"""See base class."""
|
| 393 |
+
return ["0", "1"]
|
| 394 |
+
|
| 395 |
+
def _create_examples(self, lines, set_type):
|
| 396 |
+
"""Creates examples for the training and dev sets."""
|
| 397 |
+
examples = []
|
| 398 |
+
for (i, line) in enumerate(lines):
|
| 399 |
+
if i == 0:
|
| 400 |
+
continue
|
| 401 |
+
guid = "%s-%s" % (set_type, line[0])
|
| 402 |
+
text_a = line[1]
|
| 403 |
+
text_b = line[2]
|
| 404 |
+
label = line[-1]
|
| 405 |
+
examples.append(
|
| 406 |
+
InputExample(guid=guid, text_a=text_a, text_b=text_b, label=label))
|
| 407 |
+
return examples
|
| 408 |
+
|
| 409 |
+
|
| 410 |
+
def convert_examples_to_features(examples, label_list, max_seq_length,
|
| 411 |
+
tokenizer, output_mode):
|
| 412 |
+
"""Loads a data file into a list of `InputBatch`s."""
|
| 413 |
+
|
| 414 |
+
label_map = {label : i for i, label in enumerate(label_list)}
|
| 415 |
+
|
| 416 |
+
features = []
|
| 417 |
+
for (ex_index, example) in enumerate(examples):
|
| 418 |
+
if ex_index % 10000 == 0:
|
| 419 |
+
logger.info("Writing example %d of %d" % (ex_index, len(examples)))
|
| 420 |
+
|
| 421 |
+
tokens_a = tokenizer.tokenize(example.text_a)
|
| 422 |
+
|
| 423 |
+
tokens_b = None
|
| 424 |
+
if example.text_b:
|
| 425 |
+
tokens_b = tokenizer.tokenize(example.text_b)
|
| 426 |
+
# Modifies `tokens_a` and `tokens_b` in place so that the total
|
| 427 |
+
# length is less than the specified length.
|
| 428 |
+
# Account for [CLS], [SEP], [SEP] with "- 3"
|
| 429 |
+
_truncate_seq_pair(tokens_a, tokens_b, max_seq_length - 3)
|
| 430 |
+
else:
|
| 431 |
+
# Account for [CLS] and [SEP] with "- 2"
|
| 432 |
+
if len(tokens_a) > max_seq_length - 2:
|
| 433 |
+
tokens_a = tokens_a[:(max_seq_length - 2)]
|
| 434 |
+
|
| 435 |
+
# The convention in BERT is:
|
| 436 |
+
# (a) For sequence pairs:
|
| 437 |
+
# tokens: [CLS] is this jack ##son ##ville ? [SEP] no it is not . [SEP]
|
| 438 |
+
# type_ids: 0 0 0 0 0 0 0 0 1 1 1 1 1 1
|
| 439 |
+
# (b) For single sequences:
|
| 440 |
+
# tokens: [CLS] the dog is hairy . [SEP]
|
| 441 |
+
# type_ids: 0 0 0 0 0 0 0
|
| 442 |
+
#
|
| 443 |
+
# Where "type_ids" are used to indicate whether this is the first
|
| 444 |
+
# sequence or the second sequence. The embedding vectors for `type=0` and
|
| 445 |
+
# `type=1` were learned during pre-training and are added to the wordpiece
|
| 446 |
+
# embedding vector (and position vector). This is not *strictly* necessary
|
| 447 |
+
# since the [SEP] token unambiguously separates the sequences, but it makes
|
| 448 |
+
# it easier for the model to learn the concept of sequences.
|
| 449 |
+
#
|
| 450 |
+
# For classification tasks, the first vector (corresponding to [CLS]) is
|
| 451 |
+
# used as as the "sentence vector". Note that this only makes sense because
|
| 452 |
+
# the entire model is fine-tuned.
|
| 453 |
+
tokens = ["[CLS]"] + tokens_a + ["[SEP]"]
|
| 454 |
+
segment_ids = [0] * len(tokens)
|
| 455 |
+
|
| 456 |
+
if tokens_b:
|
| 457 |
+
tokens += tokens_b + ["[SEP]"]
|
| 458 |
+
segment_ids += [1] * (len(tokens_b) + 1)
|
| 459 |
+
|
| 460 |
+
input_ids = tokenizer.convert_tokens_to_ids(tokens)
|
| 461 |
+
|
| 462 |
+
# The mask has 1 for real tokens and 0 for padding tokens. Only real
|
| 463 |
+
# tokens are attended to.
|
| 464 |
+
input_mask = [1] * len(input_ids)
|
| 465 |
+
|
| 466 |
+
# Zero-pad up to the sequence length.
|
| 467 |
+
padding = [0] * (max_seq_length - len(input_ids))
|
| 468 |
+
input_ids += padding
|
| 469 |
+
input_mask += padding
|
| 470 |
+
segment_ids += padding
|
| 471 |
+
|
| 472 |
+
assert len(input_ids) == max_seq_length
|
| 473 |
+
assert len(input_mask) == max_seq_length
|
| 474 |
+
assert len(segment_ids) == max_seq_length
|
| 475 |
+
|
| 476 |
+
if output_mode == "classification":
|
| 477 |
+
label_id = label_map[example.label]
|
| 478 |
+
elif output_mode == "regression":
|
| 479 |
+
label_id = float(example.label)
|
| 480 |
+
else:
|
| 481 |
+
raise KeyError(output_mode)
|
| 482 |
+
|
| 483 |
+
if ex_index < 5:
|
| 484 |
+
logger.info("*** Example ***")
|
| 485 |
+
logger.info("guid: %s" % (example.guid))
|
| 486 |
+
logger.info("tokens: %s" % " ".join(
|
| 487 |
+
[str(x) for x in tokens]))
|
| 488 |
+
logger.info("input_ids: %s" % " ".join([str(x) for x in input_ids]))
|
| 489 |
+
logger.info("input_mask: %s" % " ".join([str(x) for x in input_mask]))
|
| 490 |
+
logger.info(
|
| 491 |
+
"segment_ids: %s" % " ".join([str(x) for x in segment_ids]))
|
| 492 |
+
logger.info("label: %s (id = %d)" % (example.label, label_id))
|
| 493 |
+
|
| 494 |
+
features.append(
|
| 495 |
+
InputFeatures(input_ids=input_ids,
|
| 496 |
+
input_mask=input_mask,
|
| 497 |
+
segment_ids=segment_ids,
|
| 498 |
+
label_id=label_id))
|
| 499 |
+
return features
|
| 500 |
+
|
| 501 |
+
|
| 502 |
+
def _truncate_seq_pair(tokens_a, tokens_b, max_length):
|
| 503 |
+
"""Truncates a sequence pair in place to the maximum length."""
|
| 504 |
+
|
| 505 |
+
# This is a simple heuristic which will always truncate the longer sequence
|
| 506 |
+
# one token at a time. This makes more sense than truncating an equal percent
|
| 507 |
+
# of tokens from each, since if one sequence is very short then each token
|
| 508 |
+
# that's truncated likely contains more information than a longer sequence.
|
| 509 |
+
while True:
|
| 510 |
+
total_length = len(tokens_a) + len(tokens_b)
|
| 511 |
+
if total_length <= max_length:
|
| 512 |
+
break
|
| 513 |
+
if len(tokens_a) > len(tokens_b):
|
| 514 |
+
tokens_a.pop()
|
| 515 |
+
else:
|
| 516 |
+
tokens_b.pop()
|
| 517 |
+
|
| 518 |
+
|
| 519 |
+
def simple_accuracy(preds, labels):
|
| 520 |
+
return (preds == labels).mean()
|
| 521 |
+
|
| 522 |
+
|
| 523 |
+
def acc_and_f1(preds, labels):
|
| 524 |
+
acc = simple_accuracy(preds, labels)
|
| 525 |
+
f1 = f1_score(y_true=labels, y_pred=preds)
|
| 526 |
+
return {
|
| 527 |
+
"acc": acc,
|
| 528 |
+
"f1": f1,
|
| 529 |
+
"acc_and_f1": (acc + f1) / 2,
|
| 530 |
+
}
|
| 531 |
+
|
| 532 |
+
|
| 533 |
+
def pearson_and_spearman(preds, labels):
|
| 534 |
+
pearson_corr = pearsonr(preds, labels)[0]
|
| 535 |
+
spearman_corr = spearmanr(preds, labels)[0]
|
| 536 |
+
return {
|
| 537 |
+
"pearson": pearson_corr,
|
| 538 |
+
"spearmanr": spearman_corr,
|
| 539 |
+
"corr": (pearson_corr + spearman_corr) / 2,
|
| 540 |
+
}
|
| 541 |
+
|
| 542 |
+
|
| 543 |
+
def compute_metrics(task_name, preds, labels):
|
| 544 |
+
assert len(preds) == len(labels)
|
| 545 |
+
if task_name == "cola":
|
| 546 |
+
return {"mcc": matthews_corrcoef(labels, preds)}
|
| 547 |
+
elif task_name == "sst-2":
|
| 548 |
+
return {"acc": simple_accuracy(preds, labels)}
|
| 549 |
+
elif task_name == "mrpc":
|
| 550 |
+
return acc_and_f1(preds, labels)
|
| 551 |
+
elif task_name == "sts-b":
|
| 552 |
+
return pearson_and_spearman(preds, labels)
|
| 553 |
+
elif task_name == "qqp":
|
| 554 |
+
return acc_and_f1(preds, labels)
|
| 555 |
+
elif task_name == "mnli":
|
| 556 |
+
return {"acc": simple_accuracy(preds, labels)}
|
| 557 |
+
elif task_name == "mnli-mm":
|
| 558 |
+
return {"acc": simple_accuracy(preds, labels)}
|
| 559 |
+
elif task_name == "qnli":
|
| 560 |
+
return {"acc": simple_accuracy(preds, labels)}
|
| 561 |
+
elif task_name == "rte":
|
| 562 |
+
return {"acc": simple_accuracy(preds, labels)}
|
| 563 |
+
elif task_name == "wnli":
|
| 564 |
+
return {"acc": simple_accuracy(preds, labels)}
|
| 565 |
+
else:
|
| 566 |
+
raise KeyError(task_name)
|
| 567 |
+
|
| 568 |
+
|
| 569 |
+
def main():
|
| 570 |
+
parser = argparse.ArgumentParser()
|
| 571 |
+
|
| 572 |
+
## Required parameters
|
| 573 |
+
parser.add_argument("--data_dir",
|
| 574 |
+
default=None,
|
| 575 |
+
type=str,
|
| 576 |
+
required=True,
|
| 577 |
+
help="The input data dir. Should contain the .tsv files (or other data files) for the task.")
|
| 578 |
+
parser.add_argument("--bert_model", default=None, type=str, required=True,
|
| 579 |
+
help="Bert pre-trained model selected in the list: bert-base-uncased, "
|
| 580 |
+
"bert-large-uncased, bert-base-cased, bert-large-cased, bert-base-multilingual-uncased, "
|
| 581 |
+
"bert-base-multilingual-cased, bert-base-chinese.")
|
| 582 |
+
parser.add_argument("--task_name",
|
| 583 |
+
default=None,
|
| 584 |
+
type=str,
|
| 585 |
+
required=True,
|
| 586 |
+
help="The name of the task to train.")
|
| 587 |
+
parser.add_argument("--output_dir",
|
| 588 |
+
default=None,
|
| 589 |
+
type=str,
|
| 590 |
+
required=True,
|
| 591 |
+
help="The output directory where the model predictions and checkpoints will be written.")
|
| 592 |
+
|
| 593 |
+
## Other parameters
|
| 594 |
+
parser.add_argument("--cache_dir",
|
| 595 |
+
default="",
|
| 596 |
+
type=str,
|
| 597 |
+
help="Where do you want to store the pre-trained models downloaded from s3")
|
| 598 |
+
parser.add_argument("--max_seq_length",
|
| 599 |
+
default=128,
|
| 600 |
+
type=int,
|
| 601 |
+
help="The maximum total input sequence length after WordPiece tokenization. \n"
|
| 602 |
+
"Sequences longer than this will be truncated, and sequences shorter \n"
|
| 603 |
+
"than this will be padded.")
|
| 604 |
+
parser.add_argument("--do_train",
|
| 605 |
+
action='store_true',
|
| 606 |
+
help="Whether to run training.")
|
| 607 |
+
parser.add_argument("--do_eval",
|
| 608 |
+
action='store_true',
|
| 609 |
+
help="Whether to run eval on the dev set.")
|
| 610 |
+
parser.add_argument("--do_lower_case",
|
| 611 |
+
action='store_true',
|
| 612 |
+
help="Set this flag if you are using an uncased model.")
|
| 613 |
+
parser.add_argument("--train_batch_size",
|
| 614 |
+
default=32,
|
| 615 |
+
type=int,
|
| 616 |
+
help="Total batch size for training.")
|
| 617 |
+
parser.add_argument("--eval_batch_size",
|
| 618 |
+
default=8,
|
| 619 |
+
type=int,
|
| 620 |
+
help="Total batch size for eval.")
|
| 621 |
+
parser.add_argument("--learning_rate",
|
| 622 |
+
default=5e-5,
|
| 623 |
+
type=float,
|
| 624 |
+
help="The initial learning rate for Adam.")
|
| 625 |
+
parser.add_argument("--num_train_epochs",
|
| 626 |
+
default=3.0,
|
| 627 |
+
type=float,
|
| 628 |
+
help="Total number of training epochs to perform.")
|
| 629 |
+
parser.add_argument("--warmup_proportion",
|
| 630 |
+
default=0.1,
|
| 631 |
+
type=float,
|
| 632 |
+
help="Proportion of training to perform linear learning rate warmup for. "
|
| 633 |
+
"E.g., 0.1 = 10%% of training.")
|
| 634 |
+
parser.add_argument("--no_cuda",
|
| 635 |
+
action='store_true',
|
| 636 |
+
help="Whether not to use CUDA when available")
|
| 637 |
+
parser.add_argument("--local_rank",
|
| 638 |
+
type=int,
|
| 639 |
+
default=-1,
|
| 640 |
+
help="local_rank for distributed training on gpus")
|
| 641 |
+
parser.add_argument('--seed',
|
| 642 |
+
type=int,
|
| 643 |
+
default=42,
|
| 644 |
+
help="random seed for initialization")
|
| 645 |
+
parser.add_argument('--gradient_accumulation_steps',
|
| 646 |
+
type=int,
|
| 647 |
+
default=1,
|
| 648 |
+
help="Number of updates steps to accumulate before performing a backward/update pass.")
|
| 649 |
+
parser.add_argument('--fp16',
|
| 650 |
+
action='store_true',
|
| 651 |
+
help="Whether to use 16-bit float precision instead of 32-bit")
|
| 652 |
+
parser.add_argument('--loss_scale',
|
| 653 |
+
type=float, default=0,
|
| 654 |
+
help="Loss scaling to improve fp16 numeric stability. Only used when fp16 set to True.\n"
|
| 655 |
+
"0 (default value): dynamic loss scaling.\n"
|
| 656 |
+
"Positive power of 2: static loss scaling value.\n")
|
| 657 |
+
parser.add_argument('--server_ip', type=str, default='', help="Can be used for distant debugging.")
|
| 658 |
+
parser.add_argument('--server_port', type=str, default='', help="Can be used for distant debugging.")
|
| 659 |
+
args = parser.parse_args()
|
| 660 |
+
|
| 661 |
+
if args.server_ip and args.server_port:
|
| 662 |
+
# Distant debugging - see https://code.visualstudio.com/docs/python/debugging#_attach-to-a-local-script
|
| 663 |
+
import ptvsd
|
| 664 |
+
print("Waiting for debugger attach")
|
| 665 |
+
ptvsd.enable_attach(address=(args.server_ip, args.server_port), redirect_output=True)
|
| 666 |
+
ptvsd.wait_for_attach()
|
| 667 |
+
|
| 668 |
+
processors = {
|
| 669 |
+
"cola": ColaProcessor,
|
| 670 |
+
"mnli": MnliProcessor,
|
| 671 |
+
"mnli-mm": MnliMismatchedProcessor,
|
| 672 |
+
"mrpc": MrpcProcessor,
|
| 673 |
+
"sst-2": Sst2Processor,
|
| 674 |
+
"sts-b": StsbProcessor,
|
| 675 |
+
"qqp": QqpProcessor,
|
| 676 |
+
"qnli": QnliProcessor,
|
| 677 |
+
"rte": RteProcessor,
|
| 678 |
+
"wnli": WnliProcessor,
|
| 679 |
+
}
|
| 680 |
+
|
| 681 |
+
output_modes = {
|
| 682 |
+
"cola": "classification",
|
| 683 |
+
"mnli": "classification",
|
| 684 |
+
"mrpc": "classification",
|
| 685 |
+
"sst-2": "classification",
|
| 686 |
+
"sts-b": "regression",
|
| 687 |
+
"qqp": "classification",
|
| 688 |
+
"qnli": "classification",
|
| 689 |
+
"rte": "classification",
|
| 690 |
+
"wnli": "classification",
|
| 691 |
+
}
|
| 692 |
+
|
| 693 |
+
if args.local_rank == -1 or args.no_cuda:
|
| 694 |
+
device = torch.device("cuda" if torch.cuda.is_available() and not args.no_cuda else "cpu")
|
| 695 |
+
n_gpu = torch.cuda.device_count()
|
| 696 |
+
else:
|
| 697 |
+
torch.cuda.set_device(args.local_rank)
|
| 698 |
+
device = torch.device("cuda", args.local_rank)
|
| 699 |
+
n_gpu = 1
|
| 700 |
+
# Initializes the distributed backend which will take care of sychronizing nodes/GPUs
|
| 701 |
+
torch.distributed.init_process_group(backend='nccl')
|
| 702 |
+
|
| 703 |
+
logging.basicConfig(format = '%(asctime)s - %(levelname)s - %(name)s - %(message)s',
|
| 704 |
+
datefmt = '%m/%d/%Y %H:%M:%S',
|
| 705 |
+
level = logging.INFO if args.local_rank in [-1, 0] else logging.WARN)
|
| 706 |
+
|
| 707 |
+
logger.info("device: {} n_gpu: {}, distributed training: {}, 16-bits training: {}".format(
|
| 708 |
+
device, n_gpu, bool(args.local_rank != -1), args.fp16))
|
| 709 |
+
|
| 710 |
+
if args.gradient_accumulation_steps < 1:
|
| 711 |
+
raise ValueError("Invalid gradient_accumulation_steps parameter: {}, should be >= 1".format(
|
| 712 |
+
args.gradient_accumulation_steps))
|
| 713 |
+
|
| 714 |
+
args.train_batch_size = args.train_batch_size // args.gradient_accumulation_steps
|
| 715 |
+
|
| 716 |
+
random.seed(args.seed)
|
| 717 |
+
np.random.seed(args.seed)
|
| 718 |
+
torch.manual_seed(args.seed)
|
| 719 |
+
if n_gpu > 0:
|
| 720 |
+
torch.cuda.manual_seed_all(args.seed)
|
| 721 |
+
|
| 722 |
+
if not args.do_train and not args.do_eval:
|
| 723 |
+
raise ValueError("At least one of `do_train` or `do_eval` must be True.")
|
| 724 |
+
|
| 725 |
+
if os.path.exists(args.output_dir) and os.listdir(args.output_dir) and args.do_train:
|
| 726 |
+
raise ValueError("Output directory ({}) already exists and is not empty.".format(args.output_dir))
|
| 727 |
+
if not os.path.exists(args.output_dir):
|
| 728 |
+
os.makedirs(args.output_dir)
|
| 729 |
+
|
| 730 |
+
task_name = args.task_name.lower()
|
| 731 |
+
|
| 732 |
+
if task_name not in processors:
|
| 733 |
+
raise ValueError("Task not found: %s" % (task_name))
|
| 734 |
+
|
| 735 |
+
processor = processors[task_name]()
|
| 736 |
+
output_mode = output_modes[task_name]
|
| 737 |
+
|
| 738 |
+
label_list = processor.get_labels()
|
| 739 |
+
num_labels = len(label_list)
|
| 740 |
+
|
| 741 |
+
tokenizer = BertTokenizer.from_pretrained(args.bert_model, do_lower_case=args.do_lower_case)
|
| 742 |
+
|
| 743 |
+
train_examples = None
|
| 744 |
+
num_train_optimization_steps = None
|
| 745 |
+
if args.do_train:
|
| 746 |
+
train_examples = processor.get_train_examples(args.data_dir)
|
| 747 |
+
num_train_optimization_steps = int(
|
| 748 |
+
len(train_examples) / args.train_batch_size / args.gradient_accumulation_steps) * args.num_train_epochs
|
| 749 |
+
if args.local_rank != -1:
|
| 750 |
+
num_train_optimization_steps = num_train_optimization_steps // torch.distributed.get_world_size()
|
| 751 |
+
|
| 752 |
+
# Prepare model
|
| 753 |
+
cache_dir = args.cache_dir if args.cache_dir else os.path.join(str(PYTORCH_PRETRAINED_BERT_CACHE), 'distributed_{}'.format(args.local_rank))
|
| 754 |
+
model = BertForSequenceClassification.from_pretrained(args.bert_model,
|
| 755 |
+
cache_dir=cache_dir,
|
| 756 |
+
num_labels=num_labels)
|
| 757 |
+
|
| 758 |
+
### RANDOM INITIALIZATION ####
|
| 759 |
+
# config = BertConfig.from_dict({
|
| 760 |
+
# "attention_probs_dropout_prob": 0.1,
|
| 761 |
+
# "hidden_act": "gelu",
|
| 762 |
+
# "hidden_dropout_prob": 0.1,
|
| 763 |
+
# "hidden_size": 768,
|
| 764 |
+
# "initializer_range": 0.02,
|
| 765 |
+
# "intermediate_size": 3072,
|
| 766 |
+
# "max_position_embeddings": 512,
|
| 767 |
+
# "num_attention_heads": 12,
|
| 768 |
+
# "num_hidden_layers": 12,
|
| 769 |
+
# "type_vocab_size": 2,
|
| 770 |
+
# "vocab_size": 30522
|
| 771 |
+
# })
|
| 772 |
+
# model = BertForSequenceClassification(config=config, num_labels=num_labels)
|
| 773 |
+
|
| 774 |
+
|
| 775 |
+
###############################
|
| 776 |
+
|
| 777 |
+
if args.fp16:
|
| 778 |
+
model.half()
|
| 779 |
+
model.to(device)
|
| 780 |
+
if args.local_rank != -1:
|
| 781 |
+
try:
|
| 782 |
+
from apex.parallel import DistributedDataParallel as DDP
|
| 783 |
+
except ImportError:
|
| 784 |
+
raise ImportError("Please install apex from https://www.github.com/nvidia/apex to use distributed and fp16 training.")
|
| 785 |
+
|
| 786 |
+
model = DDP(model)
|
| 787 |
+
elif n_gpu > 1:
|
| 788 |
+
model = torch.nn.DataParallel(model)
|
| 789 |
+
|
| 790 |
+
# Prepare optimizer
|
| 791 |
+
param_optimizer = list(model.named_parameters())
|
| 792 |
+
no_decay = ['bias', 'LayerNorm.bias', 'LayerNorm.weight']
|
| 793 |
+
optimizer_grouped_parameters = [
|
| 794 |
+
{'params': [p for n, p in param_optimizer if not any(nd in n for nd in no_decay)], 'weight_decay': 0.01},
|
| 795 |
+
{'params': [p for n, p in param_optimizer if any(nd in n for nd in no_decay)], 'weight_decay': 0.0}
|
| 796 |
+
]
|
| 797 |
+
if args.fp16:
|
| 798 |
+
try:
|
| 799 |
+
from apex.optimizers import FP16_Optimizer
|
| 800 |
+
from apex.optimizers import FusedAdam
|
| 801 |
+
except ImportError:
|
| 802 |
+
raise ImportError("Please install apex from https://www.github.com/nvidia/apex to use distributed and fp16 training.")
|
| 803 |
+
|
| 804 |
+
optimizer = FusedAdam(optimizer_grouped_parameters,
|
| 805 |
+
lr=args.learning_rate,
|
| 806 |
+
bias_correction=False,
|
| 807 |
+
max_grad_norm=1.0)
|
| 808 |
+
if args.loss_scale == 0:
|
| 809 |
+
optimizer = FP16_Optimizer(optimizer, dynamic_loss_scale=True)
|
| 810 |
+
else:
|
| 811 |
+
optimizer = FP16_Optimizer(optimizer, static_loss_scale=args.loss_scale)
|
| 812 |
+
|
| 813 |
+
else:
|
| 814 |
+
optimizer = BertAdam(optimizer_grouped_parameters,
|
| 815 |
+
lr=args.learning_rate,
|
| 816 |
+
warmup=args.warmup_proportion,
|
| 817 |
+
t_total=num_train_optimization_steps)
|
| 818 |
+
|
| 819 |
+
global_step = 0
|
| 820 |
+
nb_tr_steps = 0
|
| 821 |
+
tr_loss = 0
|
| 822 |
+
if args.do_train:
|
| 823 |
+
train_features = convert_examples_to_features(
|
| 824 |
+
train_examples, label_list, args.max_seq_length, tokenizer, output_mode)
|
| 825 |
+
logger.info("***** Running training *****")
|
| 826 |
+
logger.info(" Num examples = %d", len(train_examples))
|
| 827 |
+
logger.info(" Batch size = %d", args.train_batch_size)
|
| 828 |
+
logger.info(" Num steps = %d", num_train_optimization_steps)
|
| 829 |
+
all_input_ids = torch.tensor([f.input_ids for f in train_features], dtype=torch.long)
|
| 830 |
+
all_input_mask = torch.tensor([f.input_mask for f in train_features], dtype=torch.long)
|
| 831 |
+
all_segment_ids = torch.tensor([f.segment_ids for f in train_features], dtype=torch.long)
|
| 832 |
+
|
| 833 |
+
if output_mode == "classification":
|
| 834 |
+
all_label_ids = torch.tensor([f.label_id for f in train_features], dtype=torch.long)
|
| 835 |
+
elif output_mode == "regression":
|
| 836 |
+
all_label_ids = torch.tensor([f.label_id for f in train_features], dtype=torch.float)
|
| 837 |
+
|
| 838 |
+
train_data = TensorDataset(all_input_ids, all_input_mask, all_segment_ids, all_label_ids)
|
| 839 |
+
if args.local_rank == -1:
|
| 840 |
+
train_sampler = RandomSampler(train_data)
|
| 841 |
+
else:
|
| 842 |
+
train_sampler = DistributedSampler(train_data)
|
| 843 |
+
train_dataloader = DataLoader(train_data, sampler=train_sampler, batch_size=args.train_batch_size)
|
| 844 |
+
|
| 845 |
+
model.train()
|
| 846 |
+
for _ in trange(int(args.num_train_epochs), desc="Epoch"):
|
| 847 |
+
tr_loss = 0
|
| 848 |
+
nb_tr_examples, nb_tr_steps = 0, 0
|
| 849 |
+
for step, batch in enumerate(tqdm(train_dataloader, desc="Iteration")):
|
| 850 |
+
batch = tuple(t.to(device) for t in batch)
|
| 851 |
+
input_ids, input_mask, segment_ids, label_ids = batch
|
| 852 |
+
|
| 853 |
+
# define a new function to compute loss values for both output_modes
|
| 854 |
+
logits, _ = model(input_ids, segment_ids, input_mask, labels=None)
|
| 855 |
+
|
| 856 |
+
if output_mode == "classification":
|
| 857 |
+
loss_fct = CrossEntropyLoss()
|
| 858 |
+
loss = loss_fct(logits.view(-1, num_labels), label_ids.view(-1))
|
| 859 |
+
elif output_mode == "regression":
|
| 860 |
+
loss_fct = MSELoss()
|
| 861 |
+
loss = loss_fct(logits.view(-1), label_ids.view(-1))
|
| 862 |
+
|
| 863 |
+
if n_gpu > 1:
|
| 864 |
+
loss = loss.mean() # mean() to average on multi-gpu.
|
| 865 |
+
if args.gradient_accumulation_steps > 1:
|
| 866 |
+
loss = loss / args.gradient_accumulation_steps
|
| 867 |
+
|
| 868 |
+
if args.fp16:
|
| 869 |
+
optimizer.backward(loss)
|
| 870 |
+
else:
|
| 871 |
+
loss.backward()
|
| 872 |
+
|
| 873 |
+
tr_loss += loss.item()
|
| 874 |
+
print(loss.item())
|
| 875 |
+
nb_tr_examples += input_ids.size(0)
|
| 876 |
+
nb_tr_steps += 1
|
| 877 |
+
if (step + 1) % args.gradient_accumulation_steps == 0:
|
| 878 |
+
if args.fp16:
|
| 879 |
+
# modify learning rate with special warm up BERT uses
|
| 880 |
+
# if args.fp16 is False, BertAdam is used that handles this automatically
|
| 881 |
+
lr_this_step = args.learning_rate * warmup_linear(global_step/num_train_optimization_steps, args.warmup_proportion)
|
| 882 |
+
for param_group in optimizer.param_groups:
|
| 883 |
+
param_group['lr'] = lr_this_step
|
| 884 |
+
optimizer.step()
|
| 885 |
+
optimizer.zero_grad()
|
| 886 |
+
global_step += 1
|
| 887 |
+
|
| 888 |
+
if args.do_train and (args.local_rank == -1 or torch.distributed.get_rank() == 0):
|
| 889 |
+
# Save a trained model, configuration and tokenizer
|
| 890 |
+
model_to_save = model.module if hasattr(model, 'module') else model # Only save the model it-self
|
| 891 |
+
|
| 892 |
+
# If we save using the predefined names, we can load using `from_pretrained`
|
| 893 |
+
output_model_file = os.path.join(args.output_dir, WEIGHTS_NAME)
|
| 894 |
+
output_config_file = os.path.join(args.output_dir, CONFIG_NAME)
|
| 895 |
+
|
| 896 |
+
torch.save(model_to_save.state_dict(), output_model_file)
|
| 897 |
+
model_to_save.config.to_json_file(output_config_file)
|
| 898 |
+
tokenizer.save_vocabulary(args.output_dir)
|
| 899 |
+
|
| 900 |
+
# Load a trained model and vocabulary that you have fine-tuned
|
| 901 |
+
model = BertForSequenceClassification.from_pretrained(args.output_dir, num_labels=num_labels)
|
| 902 |
+
tokenizer = BertTokenizer.from_pretrained(args.output_dir, do_lower_case=args.do_lower_case)
|
| 903 |
+
else:
|
| 904 |
+
model = BertForSequenceClassification.from_pretrained(args.bert_model, num_labels=num_labels)
|
| 905 |
+
model.to(device)
|
| 906 |
+
|
| 907 |
+
if args.do_eval and (args.local_rank == -1 or torch.distributed.get_rank() == 0):
|
| 908 |
+
eval_examples = processor.get_dev_examples(args.data_dir)
|
| 909 |
+
eval_features = convert_examples_to_features(
|
| 910 |
+
eval_examples, label_list, args.max_seq_length, tokenizer, output_mode)
|
| 911 |
+
logger.info("***** Running evaluation *****")
|
| 912 |
+
logger.info(" Num examples = %d", len(eval_examples))
|
| 913 |
+
logger.info(" Batch size = %d", args.eval_batch_size)
|
| 914 |
+
all_input_ids = torch.tensor([f.input_ids for f in eval_features], dtype=torch.long)
|
| 915 |
+
all_input_mask = torch.tensor([f.input_mask for f in eval_features], dtype=torch.long)
|
| 916 |
+
all_segment_ids = torch.tensor([f.segment_ids for f in eval_features], dtype=torch.long)
|
| 917 |
+
|
| 918 |
+
if output_mode == "classification":
|
| 919 |
+
all_label_ids = torch.tensor([f.label_id for f in eval_features], dtype=torch.long)
|
| 920 |
+
elif output_mode == "regression":
|
| 921 |
+
all_label_ids = torch.tensor([f.label_id for f in eval_features], dtype=torch.float)
|
| 922 |
+
|
| 923 |
+
eval_data = TensorDataset(all_input_ids, all_input_mask, all_segment_ids, all_label_ids)
|
| 924 |
+
# Run prediction for full data
|
| 925 |
+
eval_sampler = SequentialSampler(eval_data)
|
| 926 |
+
eval_dataloader = DataLoader(eval_data, sampler=eval_sampler, batch_size=args.eval_batch_size)
|
| 927 |
+
|
| 928 |
+
model.eval()
|
| 929 |
+
eval_loss = 0
|
| 930 |
+
nb_eval_steps = 0
|
| 931 |
+
preds = []
|
| 932 |
+
|
| 933 |
+
for input_ids, input_mask, segment_ids, label_ids in tqdm(eval_dataloader, desc="Evaluating"):
|
| 934 |
+
input_ids = input_ids.to(device)
|
| 935 |
+
input_mask = input_mask.to(device)
|
| 936 |
+
segment_ids = segment_ids.to(device)
|
| 937 |
+
label_ids = label_ids.to(device)
|
| 938 |
+
|
| 939 |
+
with torch.no_grad():
|
| 940 |
+
logits, attns = model(input_ids, segment_ids, input_mask, labels=None)
|
| 941 |
+
|
| 942 |
+
# create eval loss and other metric required by the task
|
| 943 |
+
if output_mode == "classification":
|
| 944 |
+
loss_fct = CrossEntropyLoss()
|
| 945 |
+
tmp_eval_loss = loss_fct(logits.view(-1, num_labels), label_ids.view(-1))
|
| 946 |
+
elif output_mode == "regression":
|
| 947 |
+
loss_fct = MSELoss()
|
| 948 |
+
tmp_eval_loss = loss_fct(logits.view(-1), label_ids.view(-1))
|
| 949 |
+
|
| 950 |
+
eval_loss += tmp_eval_loss.mean().item()
|
| 951 |
+
nb_eval_steps += 1
|
| 952 |
+
if len(preds) == 0:
|
| 953 |
+
preds.append(logits.detach().cpu().numpy())
|
| 954 |
+
else:
|
| 955 |
+
preds[0] = np.append(preds[0], logits.detach().cpu().numpy(), axis=0)
|
| 956 |
+
|
| 957 |
+
eval_loss = eval_loss / nb_eval_steps
|
| 958 |
+
preds = preds[0]
|
| 959 |
+
if output_mode == "classification":
|
| 960 |
+
preds = np.argmax(preds, axis=1)
|
| 961 |
+
elif output_mode == "regression":
|
| 962 |
+
preds = np.squeeze(preds)
|
| 963 |
+
result = compute_metrics(task_name, preds, all_label_ids.numpy())
|
| 964 |
+
loss = tr_loss/nb_tr_steps if args.do_train else None
|
| 965 |
+
|
| 966 |
+
result['eval_loss'] = eval_loss
|
| 967 |
+
result['global_step'] = global_step
|
| 968 |
+
result['loss'] = loss
|
| 969 |
+
|
| 970 |
+
output_eval_file = os.path.join(args.output_dir, "eval_results.txt")
|
| 971 |
+
with open(output_eval_file, "w") as writer:
|
| 972 |
+
logger.info("***** Eval results *****")
|
| 973 |
+
for key in sorted(result.keys()):
|
| 974 |
+
logger.info(" %s = %s", key, str(result[key]))
|
| 975 |
+
writer.write("%s = %s\n" % (key, str(result[key])))
|
| 976 |
+
|
| 977 |
+
# hack for MNLI-MM
|
| 978 |
+
if task_name == "mnli":
|
| 979 |
+
task_name = "mnli-mm"
|
| 980 |
+
processor = processors[task_name]()
|
| 981 |
+
|
| 982 |
+
if os.path.exists(args.output_dir + '-MM') and os.listdir(args.output_dir + '-MM') and args.do_train:
|
| 983 |
+
raise ValueError("Output directory ({}) already exists and is not empty.".format(args.output_dir))
|
| 984 |
+
if not os.path.exists(args.output_dir + '-MM'):
|
| 985 |
+
os.makedirs(args.output_dir + '-MM')
|
| 986 |
+
|
| 987 |
+
eval_examples = processor.get_dev_examples(args.data_dir)
|
| 988 |
+
eval_features = convert_examples_to_features(
|
| 989 |
+
eval_examples, label_list, args.max_seq_length, tokenizer, output_mode)
|
| 990 |
+
logger.info("***** Running evaluation *****")
|
| 991 |
+
logger.info(" Num examples = %d", len(eval_examples))
|
| 992 |
+
logger.info(" Batch size = %d", args.eval_batch_size)
|
| 993 |
+
all_input_ids = torch.tensor([f.input_ids for f in eval_features], dtype=torch.long)
|
| 994 |
+
all_input_mask = torch.tensor([f.input_mask for f in eval_features], dtype=torch.long)
|
| 995 |
+
all_segment_ids = torch.tensor([f.segment_ids for f in eval_features], dtype=torch.long)
|
| 996 |
+
all_label_ids = torch.tensor([f.label_id for f in eval_features], dtype=torch.long)
|
| 997 |
+
|
| 998 |
+
eval_data = TensorDataset(all_input_ids, all_input_mask, all_segment_ids, all_label_ids)
|
| 999 |
+
# Run prediction for full data
|
| 1000 |
+
eval_sampler = SequentialSampler(eval_data)
|
| 1001 |
+
eval_dataloader = DataLoader(eval_data, sampler=eval_sampler, batch_size=args.eval_batch_size)
|
| 1002 |
+
|
| 1003 |
+
model.eval()
|
| 1004 |
+
eval_loss = 0
|
| 1005 |
+
nb_eval_steps = 0
|
| 1006 |
+
preds = []
|
| 1007 |
+
|
| 1008 |
+
for input_ids, input_mask, segment_ids, label_ids in tqdm(eval_dataloader, desc="Evaluating"):
|
| 1009 |
+
input_ids = input_ids.to(device)
|
| 1010 |
+
input_mask = input_mask.to(device)
|
| 1011 |
+
segment_ids = segment_ids.to(device)
|
| 1012 |
+
label_ids = label_ids.to(device)
|
| 1013 |
+
|
| 1014 |
+
with torch.no_grad():
|
| 1015 |
+
logits = model(input_ids, segment_ids, input_mask, labels=None)
|
| 1016 |
+
|
| 1017 |
+
loss_fct = CrossEntropyLoss()
|
| 1018 |
+
tmp_eval_loss = loss_fct(logits.view(-1, num_labels), label_ids.view(-1))
|
| 1019 |
+
|
| 1020 |
+
eval_loss += tmp_eval_loss.mean().item()
|
| 1021 |
+
nb_eval_steps += 1
|
| 1022 |
+
if len(preds) == 0:
|
| 1023 |
+
preds.append(logits.detach().cpu().numpy())
|
| 1024 |
+
else:
|
| 1025 |
+
preds[0] = np.append(
|
| 1026 |
+
preds[0], logits.detach().cpu().numpy(), axis=0)
|
| 1027 |
+
|
| 1028 |
+
eval_loss = eval_loss / nb_eval_steps
|
| 1029 |
+
preds = preds[0]
|
| 1030 |
+
preds = np.argmax(preds, axis=1)
|
| 1031 |
+
result = compute_metrics(task_name, preds, all_label_ids.numpy())
|
| 1032 |
+
loss = tr_loss/nb_tr_steps if args.do_train else None
|
| 1033 |
+
|
| 1034 |
+
result['eval_loss'] = eval_loss
|
| 1035 |
+
result['global_step'] = global_step
|
| 1036 |
+
result['loss'] = loss
|
| 1037 |
+
|
| 1038 |
+
output_eval_file = os.path.join(args.output_dir + '-MM', "eval_results.txt")
|
| 1039 |
+
with open(output_eval_file, "w") as writer:
|
| 1040 |
+
logger.info("***** Eval results *****")
|
| 1041 |
+
for key in sorted(result.keys()):
|
| 1042 |
+
logger.info(" %s = %s", key, str(result[key]))
|
| 1043 |
+
writer.write("%s = %s\n" % (key, str(result[key])))
|
| 1044 |
+
|
| 1045 |
+
|
| 1046 |
+
if __name__ == "__main__":
|
| 1047 |
+
main()
|
dark-secrets-of-BERT-master/dark-secrets-of-BERT-master/examples/run_gpt2.py
ADDED
|
@@ -0,0 +1,133 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
#!/usr/bin/env python3
|
| 2 |
+
|
| 3 |
+
import argparse
|
| 4 |
+
import logging
|
| 5 |
+
from tqdm import trange
|
| 6 |
+
|
| 7 |
+
import torch
|
| 8 |
+
import torch.nn.functional as F
|
| 9 |
+
import numpy as np
|
| 10 |
+
|
| 11 |
+
from pytorch_pretrained_bert import GPT2LMHeadModel, GPT2Tokenizer
|
| 12 |
+
|
| 13 |
+
logging.basicConfig(format = '%(asctime)s - %(levelname)s - %(name)s - %(message)s',
|
| 14 |
+
datefmt = '%m/%d/%Y %H:%M:%S',
|
| 15 |
+
level = logging.INFO)
|
| 16 |
+
logger = logging.getLogger(__name__)
|
| 17 |
+
|
| 18 |
+
def top_k_logits(logits, k):
|
| 19 |
+
"""
|
| 20 |
+
Masks everything but the k top entries as -infinity (1e10).
|
| 21 |
+
Used to mask logits such that e^-infinity -> 0 won't contribute to the
|
| 22 |
+
sum of the denominator.
|
| 23 |
+
"""
|
| 24 |
+
if k == 0:
|
| 25 |
+
return logits
|
| 26 |
+
else:
|
| 27 |
+
values = torch.topk(logits, k)[0]
|
| 28 |
+
batch_mins = values[:, -1].view(-1, 1).expand_as(logits)
|
| 29 |
+
return torch.where(logits < batch_mins, torch.ones_like(logits) * -1e10, logits)
|
| 30 |
+
|
| 31 |
+
def sample_sequence(model, length, start_token=None, batch_size=None, context=None, temperature=1, top_k=0, device='cuda', sample=True):
|
| 32 |
+
if start_token is None:
|
| 33 |
+
assert context is not None, 'Specify exactly one of start_token and context!'
|
| 34 |
+
context = torch.tensor(context, device=device, dtype=torch.long).unsqueeze(0).repeat(batch_size, 1)
|
| 35 |
+
else:
|
| 36 |
+
assert context is None, 'Specify exactly one of start_token and context!'
|
| 37 |
+
context = torch.full((batch_size, 1), start_token, device=device, dtype=torch.long)
|
| 38 |
+
prev = context
|
| 39 |
+
output = context
|
| 40 |
+
past = None
|
| 41 |
+
with torch.no_grad():
|
| 42 |
+
for i in trange(length):
|
| 43 |
+
logits, past = model(prev, past=past)
|
| 44 |
+
logits = logits[:, -1, :] / temperature
|
| 45 |
+
logits = top_k_logits(logits, k=top_k)
|
| 46 |
+
log_probs = F.softmax(logits, dim=-1)
|
| 47 |
+
if sample:
|
| 48 |
+
prev = torch.multinomial(log_probs, num_samples=1)
|
| 49 |
+
else:
|
| 50 |
+
_, prev = torch.topk(log_probs, k=1, dim=-1)
|
| 51 |
+
output = torch.cat((output, prev), dim=1)
|
| 52 |
+
return output
|
| 53 |
+
|
| 54 |
+
def run_model():
|
| 55 |
+
parser = argparse.ArgumentParser()
|
| 56 |
+
parser.add_argument('--model_name_or_path', type=str, default='gpt2', help='pretrained model name or path to local checkpoint')
|
| 57 |
+
parser.add_argument("--seed", type=int, default=0)
|
| 58 |
+
parser.add_argument("--nsamples", type=int, default=1)
|
| 59 |
+
parser.add_argument("--batch_size", type=int, default=-1)
|
| 60 |
+
parser.add_argument("--length", type=int, default=-1)
|
| 61 |
+
parser.add_argument("--temperature", type=float, default=1.0)
|
| 62 |
+
parser.add_argument("--top_k", type=int, default=0)
|
| 63 |
+
parser.add_argument('--unconditional', action='store_true', help='If true, unconditional generation.')
|
| 64 |
+
args = parser.parse_args()
|
| 65 |
+
print(args)
|
| 66 |
+
|
| 67 |
+
if args.batch_size == -1:
|
| 68 |
+
args.batch_size = 1
|
| 69 |
+
assert args.nsamples % args.batch_size == 0
|
| 70 |
+
|
| 71 |
+
np.random.seed(args.seed)
|
| 72 |
+
torch.random.manual_seed(args.seed)
|
| 73 |
+
torch.cuda.manual_seed(args.seed)
|
| 74 |
+
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
|
| 75 |
+
|
| 76 |
+
enc = GPT2Tokenizer.from_pretrained(args.model_name_or_path)
|
| 77 |
+
model = GPT2LMHeadModel.from_pretrained(args.model_name_or_path)
|
| 78 |
+
model.to(device)
|
| 79 |
+
model.eval()
|
| 80 |
+
|
| 81 |
+
if args.length == -1:
|
| 82 |
+
args.length = model.config.n_ctx // 2
|
| 83 |
+
elif args.length > model.config.n_ctx:
|
| 84 |
+
raise ValueError("Can't get samples longer than window size: %s" % model.config.n_ctx)
|
| 85 |
+
|
| 86 |
+
while True:
|
| 87 |
+
context_tokens = []
|
| 88 |
+
if not args.unconditional:
|
| 89 |
+
raw_text = input("Model prompt >>> ")
|
| 90 |
+
while not raw_text:
|
| 91 |
+
print('Prompt should not be empty!')
|
| 92 |
+
raw_text = input("Model prompt >>> ")
|
| 93 |
+
context_tokens = enc.encode(raw_text)
|
| 94 |
+
generated = 0
|
| 95 |
+
for _ in range(args.nsamples // args.batch_size):
|
| 96 |
+
out = sample_sequence(
|
| 97 |
+
model=model, length=args.length,
|
| 98 |
+
context=context_tokens,
|
| 99 |
+
start_token=None,
|
| 100 |
+
batch_size=args.batch_size,
|
| 101 |
+
temperature=args.temperature, top_k=args.top_k, device=device
|
| 102 |
+
)
|
| 103 |
+
out = out[:, len(context_tokens):].tolist()
|
| 104 |
+
for i in range(args.batch_size):
|
| 105 |
+
generated += 1
|
| 106 |
+
text = enc.decode(out[i])
|
| 107 |
+
print("=" * 40 + " SAMPLE " + str(generated) + " " + "=" * 40)
|
| 108 |
+
print(text)
|
| 109 |
+
print("=" * 80)
|
| 110 |
+
if args.unconditional:
|
| 111 |
+
generated = 0
|
| 112 |
+
for _ in range(args.nsamples // args.batch_size):
|
| 113 |
+
out = sample_sequence(
|
| 114 |
+
model=model, length=args.length,
|
| 115 |
+
context=None,
|
| 116 |
+
start_token=enc.encoder['<|endoftext|>'],
|
| 117 |
+
batch_size=args.batch_size,
|
| 118 |
+
temperature=args.temperature, top_k=args.top_k, device=device
|
| 119 |
+
)
|
| 120 |
+
out = out[:,1:].tolist()
|
| 121 |
+
for i in range(args.batch_size):
|
| 122 |
+
generated += 1
|
| 123 |
+
text = enc.decode(out[i])
|
| 124 |
+
print("=" * 40 + " SAMPLE " + str(generated) + " " + "=" * 40)
|
| 125 |
+
print(text)
|
| 126 |
+
print("=" * 80)
|
| 127 |
+
if args.unconditional:
|
| 128 |
+
break
|
| 129 |
+
|
| 130 |
+
if __name__ == '__main__':
|
| 131 |
+
run_model()
|
| 132 |
+
|
| 133 |
+
|
dark-secrets-of-BERT-master/dark-secrets-of-BERT-master/examples/run_openai_gpt.py
ADDED
|
@@ -0,0 +1,274 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# coding=utf-8
|
| 2 |
+
# Copyright 2018 Google AI, Google Brain and Carnegie Mellon University Authors and the HuggingFace Inc. team.
|
| 3 |
+
# Copyright (c) 2018, NVIDIA CORPORATION. All rights reserved.
|
| 4 |
+
#
|
| 5 |
+
# Licensed under the Apache License, Version 2.0 (the "License");
|
| 6 |
+
# you may not use this file except in compliance with the License.
|
| 7 |
+
# You may obtain a copy of the License at
|
| 8 |
+
#
|
| 9 |
+
# http://www.apache.org/licenses/LICENSE-2.0
|
| 10 |
+
#
|
| 11 |
+
# Unless required by applicable law or agreed to in writing, software
|
| 12 |
+
# distributed under the License is distributed on an "AS IS" BASIS,
|
| 13 |
+
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
| 14 |
+
# See the License for the specific language governing permissions and
|
| 15 |
+
# limitations under the License.
|
| 16 |
+
""" OpenAI GPT model fine-tuning script.
|
| 17 |
+
Adapted from https://github.com/huggingface/pytorch-openai-transformer-lm/blob/master/train.py
|
| 18 |
+
It self adapted from https://github.com/openai/finetune-transformer-lm/blob/master/train.py
|
| 19 |
+
|
| 20 |
+
This script with default values fine-tunes and evaluate a pretrained OpenAI GPT on the RocStories dataset:
|
| 21 |
+
python run_openai_gpt.py \
|
| 22 |
+
--model_name openai-gpt \
|
| 23 |
+
--do_train \
|
| 24 |
+
--do_eval \
|
| 25 |
+
--train_dataset $ROC_STORIES_DIR/cloze_test_val__spring2016\ -\ cloze_test_ALL_val.csv \
|
| 26 |
+
--eval_dataset $ROC_STORIES_DIR/cloze_test_test__spring2016\ -\ cloze_test_ALL_test.csv \
|
| 27 |
+
--output_dir ../log \
|
| 28 |
+
--train_batch_size 16 \
|
| 29 |
+
"""
|
| 30 |
+
import argparse
|
| 31 |
+
import os
|
| 32 |
+
import csv
|
| 33 |
+
import random
|
| 34 |
+
import logging
|
| 35 |
+
from tqdm import tqdm, trange
|
| 36 |
+
|
| 37 |
+
import numpy as np
|
| 38 |
+
import torch
|
| 39 |
+
from torch.utils.data import (DataLoader, RandomSampler, SequentialSampler,
|
| 40 |
+
TensorDataset)
|
| 41 |
+
|
| 42 |
+
from pytorch_pretrained_bert import (OpenAIGPTDoubleHeadsModel, OpenAIGPTTokenizer,
|
| 43 |
+
OpenAIAdam, cached_path, WEIGHTS_NAME, CONFIG_NAME)
|
| 44 |
+
|
| 45 |
+
ROCSTORIES_URL = "https://s3.amazonaws.com/datasets.huggingface.co/ROCStories.tar.gz"
|
| 46 |
+
|
| 47 |
+
logging.basicConfig(format = '%(asctime)s - %(levelname)s - %(name)s - %(message)s',
|
| 48 |
+
datefmt = '%m/%d/%Y %H:%M:%S',
|
| 49 |
+
level = logging.INFO)
|
| 50 |
+
logger = logging.getLogger(__name__)
|
| 51 |
+
|
| 52 |
+
def accuracy(out, labels):
|
| 53 |
+
outputs = np.argmax(out, axis=1)
|
| 54 |
+
return np.sum(outputs == labels)
|
| 55 |
+
|
| 56 |
+
def load_rocstories_dataset(dataset_path):
|
| 57 |
+
""" Output a list of tuples(story, 1st continuation, 2nd continuation, label) """
|
| 58 |
+
with open(dataset_path, encoding='utf_8') as f:
|
| 59 |
+
f = csv.reader(f)
|
| 60 |
+
output = []
|
| 61 |
+
next(f) # skip the first line
|
| 62 |
+
for line in tqdm(f):
|
| 63 |
+
output.append((' '.join(line[1:5]), line[5], line[6], int(line[-1])-1))
|
| 64 |
+
return output
|
| 65 |
+
|
| 66 |
+
def pre_process_datasets(encoded_datasets, input_len, cap_length, start_token, delimiter_token, clf_token):
|
| 67 |
+
""" Pre-process datasets containing lists of tuples(story, 1st continuation, 2nd continuation, label)
|
| 68 |
+
|
| 69 |
+
To Transformer inputs of shape (n_batch, n_alternative, length) comprising for each batch, continuation:
|
| 70 |
+
input_ids[batch, alternative, :] = [start_token] + story[:cap_length] + [delimiter_token] + cont1[:cap_length] + [clf_token]
|
| 71 |
+
"""
|
| 72 |
+
tensor_datasets = []
|
| 73 |
+
for dataset in encoded_datasets:
|
| 74 |
+
n_batch = len(dataset)
|
| 75 |
+
input_ids = np.zeros((n_batch, 2, input_len), dtype=np.int64)
|
| 76 |
+
mc_token_ids = np.zeros((n_batch, 2), dtype=np.int64)
|
| 77 |
+
lm_labels = np.full((n_batch, 2, input_len), fill_value=-1, dtype=np.int64)
|
| 78 |
+
mc_labels = np.zeros((n_batch,), dtype=np.int64)
|
| 79 |
+
for i, (story, cont1, cont2, mc_label), in enumerate(dataset):
|
| 80 |
+
with_cont1 = [start_token] + story[:cap_length] + [delimiter_token] + cont1[:cap_length] + [clf_token]
|
| 81 |
+
with_cont2 = [start_token] + story[:cap_length] + [delimiter_token] + cont2[:cap_length] + [clf_token]
|
| 82 |
+
input_ids[i, 0, :len(with_cont1)] = with_cont1
|
| 83 |
+
input_ids[i, 1, :len(with_cont2)] = with_cont2
|
| 84 |
+
mc_token_ids[i, 0] = len(with_cont1) - 1
|
| 85 |
+
mc_token_ids[i, 1] = len(with_cont2) - 1
|
| 86 |
+
lm_labels[i, 0, :len(with_cont1)-1] = with_cont1[1:]
|
| 87 |
+
lm_labels[i, 1, :len(with_cont2)-1] = with_cont2[1:]
|
| 88 |
+
mc_labels[i] = mc_label
|
| 89 |
+
all_inputs = (input_ids, mc_token_ids, lm_labels, mc_labels)
|
| 90 |
+
tensor_datasets.append(tuple(torch.tensor(t) for t in all_inputs))
|
| 91 |
+
return tensor_datasets
|
| 92 |
+
|
| 93 |
+
def main():
|
| 94 |
+
parser = argparse.ArgumentParser()
|
| 95 |
+
parser.add_argument('--model_name', type=str, default='openai-gpt',
|
| 96 |
+
help='pretrained model name')
|
| 97 |
+
parser.add_argument("--do_train", action='store_true', help="Whether to run training.")
|
| 98 |
+
parser.add_argument("--do_eval", action='store_true', help="Whether to run eval on the dev set.")
|
| 99 |
+
parser.add_argument("--output_dir", default=None, type=str, required=True,
|
| 100 |
+
help="The output directory where the model predictions and checkpoints will be written.")
|
| 101 |
+
parser.add_argument('--train_dataset', type=str, default='')
|
| 102 |
+
parser.add_argument('--eval_dataset', type=str, default='')
|
| 103 |
+
parser.add_argument('--seed', type=int, default=42)
|
| 104 |
+
parser.add_argument('--num_train_epochs', type=int, default=3)
|
| 105 |
+
parser.add_argument('--train_batch_size', type=int, default=8)
|
| 106 |
+
parser.add_argument('--eval_batch_size', type=int, default=16)
|
| 107 |
+
parser.add_argument('--max_grad_norm', type=int, default=1)
|
| 108 |
+
parser.add_argument('--learning_rate', type=float, default=6.25e-5)
|
| 109 |
+
parser.add_argument('--warmup_proportion', type=float, default=0.002)
|
| 110 |
+
parser.add_argument('--lr_schedule', type=str, default='warmup_linear')
|
| 111 |
+
parser.add_argument('--weight_decay', type=float, default=0.01)
|
| 112 |
+
parser.add_argument('--lm_coef', type=float, default=0.9)
|
| 113 |
+
parser.add_argument('--n_valid', type=int, default=374)
|
| 114 |
+
|
| 115 |
+
parser.add_argument('--server_ip', type=str, default='', help="Can be used for distant debugging.")
|
| 116 |
+
parser.add_argument('--server_port', type=str, default='', help="Can be used for distant debugging.")
|
| 117 |
+
args = parser.parse_args()
|
| 118 |
+
print(args)
|
| 119 |
+
|
| 120 |
+
if args.server_ip and args.server_port:
|
| 121 |
+
# Distant debugging - see https://code.visualstudio.com/docs/python/debugging#_attach-to-a-local-script
|
| 122 |
+
import ptvsd
|
| 123 |
+
print("Waiting for debugger attach")
|
| 124 |
+
ptvsd.enable_attach(address=(args.server_ip, args.server_port), redirect_output=True)
|
| 125 |
+
ptvsd.wait_for_attach()
|
| 126 |
+
|
| 127 |
+
random.seed(args.seed)
|
| 128 |
+
np.random.seed(args.seed)
|
| 129 |
+
torch.manual_seed(args.seed)
|
| 130 |
+
torch.cuda.manual_seed_all(args.seed)
|
| 131 |
+
|
| 132 |
+
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
|
| 133 |
+
n_gpu = torch.cuda.device_count()
|
| 134 |
+
logger.info("device: {}, n_gpu {}".format(device, n_gpu))
|
| 135 |
+
|
| 136 |
+
if not args.do_train and not args.do_eval:
|
| 137 |
+
raise ValueError("At least one of `do_train` or `do_eval` must be True.")
|
| 138 |
+
|
| 139 |
+
if not os.path.exists(args.output_dir):
|
| 140 |
+
os.makedirs(args.output_dir)
|
| 141 |
+
|
| 142 |
+
# Load tokenizer and model
|
| 143 |
+
# This loading functions also add new tokens and embeddings called `special tokens`
|
| 144 |
+
# These new embeddings will be fine-tuned on the RocStories dataset
|
| 145 |
+
special_tokens = ['_start_', '_delimiter_', '_classify_']
|
| 146 |
+
tokenizer = OpenAIGPTTokenizer.from_pretrained(args.model_name, special_tokens=special_tokens)
|
| 147 |
+
special_tokens_ids = list(tokenizer.convert_tokens_to_ids(token) for token in special_tokens)
|
| 148 |
+
model = OpenAIGPTDoubleHeadsModel.from_pretrained(args.model_name, num_special_tokens=len(special_tokens))
|
| 149 |
+
model.to(device)
|
| 150 |
+
|
| 151 |
+
# Load and encode the datasets
|
| 152 |
+
if not args.train_dataset and not args.eval_dataset:
|
| 153 |
+
roc_stories = cached_path(ROCSTORIES_URL)
|
| 154 |
+
def tokenize_and_encode(obj):
|
| 155 |
+
""" Tokenize and encode a nested object """
|
| 156 |
+
if isinstance(obj, str):
|
| 157 |
+
return tokenizer.convert_tokens_to_ids(tokenizer.tokenize(obj))
|
| 158 |
+
elif isinstance(obj, int):
|
| 159 |
+
return obj
|
| 160 |
+
return list(tokenize_and_encode(o) for o in obj)
|
| 161 |
+
logger.info("Encoding dataset...")
|
| 162 |
+
train_dataset = load_rocstories_dataset(args.train_dataset)
|
| 163 |
+
eval_dataset = load_rocstories_dataset(args.eval_dataset)
|
| 164 |
+
datasets = (train_dataset, eval_dataset)
|
| 165 |
+
encoded_datasets = tokenize_and_encode(datasets)
|
| 166 |
+
|
| 167 |
+
# Compute the max input length for the Transformer
|
| 168 |
+
max_length = model.config.n_positions // 2 - 2
|
| 169 |
+
input_length = max(len(story[:max_length]) + max(len(cont1[:max_length]), len(cont2[:max_length])) + 3 \
|
| 170 |
+
for dataset in encoded_datasets for story, cont1, cont2, _ in dataset)
|
| 171 |
+
input_length = min(input_length, model.config.n_positions) # Max size of input for the pre-trained model
|
| 172 |
+
|
| 173 |
+
# Prepare inputs tensors and dataloaders
|
| 174 |
+
tensor_datasets = pre_process_datasets(encoded_datasets, input_length, max_length, *special_tokens_ids)
|
| 175 |
+
train_tensor_dataset, eval_tensor_dataset = tensor_datasets[0], tensor_datasets[1]
|
| 176 |
+
|
| 177 |
+
train_data = TensorDataset(*train_tensor_dataset)
|
| 178 |
+
train_sampler = RandomSampler(train_data)
|
| 179 |
+
train_dataloader = DataLoader(train_data, sampler=train_sampler, batch_size=args.train_batch_size)
|
| 180 |
+
|
| 181 |
+
eval_data = TensorDataset(*eval_tensor_dataset)
|
| 182 |
+
eval_sampler = SequentialSampler(eval_data)
|
| 183 |
+
eval_dataloader = DataLoader(eval_data, sampler=eval_sampler, batch_size=args.eval_batch_size)
|
| 184 |
+
|
| 185 |
+
# Prepare optimizer
|
| 186 |
+
param_optimizer = list(model.named_parameters())
|
| 187 |
+
no_decay = ['bias', 'LayerNorm.bias', 'LayerNorm.weight']
|
| 188 |
+
optimizer_grouped_parameters = [
|
| 189 |
+
{'params': [p for n, p in param_optimizer if not any(nd in n for nd in no_decay)], 'weight_decay': 0.01},
|
| 190 |
+
{'params': [p for n, p in param_optimizer if any(nd in n for nd in no_decay)], 'weight_decay': 0.0}
|
| 191 |
+
]
|
| 192 |
+
num_train_optimization_steps = len(train_data) * args.num_train_epochs // args.train_batch_size
|
| 193 |
+
optimizer = OpenAIAdam(optimizer_grouped_parameters,
|
| 194 |
+
lr=args.learning_rate,
|
| 195 |
+
warmup=args.warmup_proportion,
|
| 196 |
+
max_grad_norm=args.max_grad_norm,
|
| 197 |
+
weight_decay=args.weight_decay,
|
| 198 |
+
t_total=num_train_optimization_steps)
|
| 199 |
+
|
| 200 |
+
if args.do_train:
|
| 201 |
+
nb_tr_steps, tr_loss, exp_average_loss = 0, 0, None
|
| 202 |
+
model.train()
|
| 203 |
+
for _ in trange(int(args.num_train_epochs), desc="Epoch"):
|
| 204 |
+
tr_loss = 0
|
| 205 |
+
nb_tr_steps = 0
|
| 206 |
+
tqdm_bar = tqdm(train_dataloader, desc="Training")
|
| 207 |
+
for step, batch in enumerate(tqdm_bar):
|
| 208 |
+
batch = tuple(t.to(device) for t in batch)
|
| 209 |
+
input_ids, mc_token_ids, lm_labels, mc_labels = batch
|
| 210 |
+
losses = model(input_ids, mc_token_ids, lm_labels, mc_labels)
|
| 211 |
+
loss = args.lm_coef * losses[0] + losses[1]
|
| 212 |
+
loss.backward()
|
| 213 |
+
optimizer.step()
|
| 214 |
+
optimizer.zero_grad()
|
| 215 |
+
tr_loss += loss.item()
|
| 216 |
+
exp_average_loss = loss.item() if exp_average_loss is None else 0.7*exp_average_loss+0.3*loss.item()
|
| 217 |
+
nb_tr_steps += 1
|
| 218 |
+
tqdm_bar.desc = "Training loss: {:.2e} lr: {:.2e}".format(exp_average_loss, optimizer.get_lr()[0])
|
| 219 |
+
|
| 220 |
+
# Save a trained model
|
| 221 |
+
if args.do_train:
|
| 222 |
+
# Save a trained model, configuration and tokenizer
|
| 223 |
+
model_to_save = model.module if hasattr(model, 'module') else model # Only save the model it-self
|
| 224 |
+
|
| 225 |
+
# If we save using the predefined names, we can load using `from_pretrained`
|
| 226 |
+
output_model_file = os.path.join(args.output_dir, WEIGHTS_NAME)
|
| 227 |
+
output_config_file = os.path.join(args.output_dir, CONFIG_NAME)
|
| 228 |
+
|
| 229 |
+
torch.save(model_to_save.state_dict(), output_model_file)
|
| 230 |
+
model_to_save.config.to_json_file(output_config_file)
|
| 231 |
+
tokenizer.save_vocabulary(args.output_dir)
|
| 232 |
+
|
| 233 |
+
# Load a trained model and vocabulary that you have fine-tuned
|
| 234 |
+
model = OpenAIGPTDoubleHeadsModel.from_pretrained(args.output_dir)
|
| 235 |
+
tokenizer = OpenAIGPTTokenizer.from_pretrained(args.output_dir)
|
| 236 |
+
model.to(device)
|
| 237 |
+
|
| 238 |
+
if args.do_eval:
|
| 239 |
+
model.eval()
|
| 240 |
+
eval_loss, eval_accuracy = 0, 0
|
| 241 |
+
nb_eval_steps, nb_eval_examples = 0, 0
|
| 242 |
+
for batch in tqdm(eval_dataloader, desc="Evaluating"):
|
| 243 |
+
batch = tuple(t.to(device) for t in batch)
|
| 244 |
+
input_ids, mc_token_ids, lm_labels, mc_labels = batch
|
| 245 |
+
with torch.no_grad():
|
| 246 |
+
_, mc_loss = model(input_ids, mc_token_ids, lm_labels, mc_labels)
|
| 247 |
+
_, mc_logits = model(input_ids, mc_token_ids)
|
| 248 |
+
|
| 249 |
+
mc_logits = mc_logits.detach().cpu().numpy()
|
| 250 |
+
mc_labels = mc_labels.to('cpu').numpy()
|
| 251 |
+
tmp_eval_accuracy = accuracy(mc_logits, mc_labels)
|
| 252 |
+
|
| 253 |
+
eval_loss += mc_loss.mean().item()
|
| 254 |
+
eval_accuracy += tmp_eval_accuracy
|
| 255 |
+
|
| 256 |
+
nb_eval_examples += input_ids.size(0)
|
| 257 |
+
nb_eval_steps += 1
|
| 258 |
+
|
| 259 |
+
eval_loss = eval_loss / nb_eval_steps
|
| 260 |
+
eval_accuracy = eval_accuracy / nb_eval_examples
|
| 261 |
+
train_loss = tr_loss/nb_tr_steps if args.do_train else None
|
| 262 |
+
result = {'eval_loss': eval_loss,
|
| 263 |
+
'eval_accuracy': eval_accuracy,
|
| 264 |
+
'train_loss': train_loss}
|
| 265 |
+
|
| 266 |
+
output_eval_file = os.path.join(args.output_dir, "eval_results.txt")
|
| 267 |
+
with open(output_eval_file, "w") as writer:
|
| 268 |
+
logger.info("***** Eval results *****")
|
| 269 |
+
for key in sorted(result.keys()):
|
| 270 |
+
logger.info(" %s = %s", key, str(result[key]))
|
| 271 |
+
writer.write("%s = %s\n" % (key, str(result[key])))
|
| 272 |
+
|
| 273 |
+
if __name__ == '__main__':
|
| 274 |
+
main()
|
dark-secrets-of-BERT-master/dark-secrets-of-BERT-master/examples/run_squad.py
ADDED
|
@@ -0,0 +1,1098 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# coding=utf-8
|
| 2 |
+
# Copyright 2018 The Google AI Language Team Authors and The HuggingFace Inc. team.
|
| 3 |
+
# Copyright (c) 2018, NVIDIA CORPORATION. All rights reserved.
|
| 4 |
+
#
|
| 5 |
+
# Licensed under the Apache License, Version 2.0 (the "License");
|
| 6 |
+
# you may not use this file except in compliance with the License.
|
| 7 |
+
# You may obtain a copy of the License at
|
| 8 |
+
#
|
| 9 |
+
# http://www.apache.org/licenses/LICENSE-2.0
|
| 10 |
+
#
|
| 11 |
+
# Unless required by applicable law or agreed to in writing, software
|
| 12 |
+
# distributed under the License is distributed on an "AS IS" BASIS,
|
| 13 |
+
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
| 14 |
+
# See the License for the specific language governing permissions and
|
| 15 |
+
# limitations under the License.
|
| 16 |
+
"""Run BERT on SQuAD."""
|
| 17 |
+
|
| 18 |
+
from __future__ import absolute_import, division, print_function
|
| 19 |
+
|
| 20 |
+
import argparse
|
| 21 |
+
import collections
|
| 22 |
+
import json
|
| 23 |
+
import logging
|
| 24 |
+
import math
|
| 25 |
+
import os
|
| 26 |
+
import random
|
| 27 |
+
import sys
|
| 28 |
+
from io import open
|
| 29 |
+
|
| 30 |
+
import numpy as np
|
| 31 |
+
import torch
|
| 32 |
+
from torch.utils.data import (DataLoader, RandomSampler, SequentialSampler,
|
| 33 |
+
TensorDataset)
|
| 34 |
+
from torch.utils.data.distributed import DistributedSampler
|
| 35 |
+
from tqdm import tqdm, trange
|
| 36 |
+
|
| 37 |
+
from pytorch_pretrained_bert.file_utils import PYTORCH_PRETRAINED_BERT_CACHE, WEIGHTS_NAME, CONFIG_NAME
|
| 38 |
+
from pytorch_pretrained_bert.modeling import BertForQuestionAnswering, BertConfig
|
| 39 |
+
from pytorch_pretrained_bert.optimization import BertAdam, warmup_linear
|
| 40 |
+
from pytorch_pretrained_bert.tokenization import (BasicTokenizer,
|
| 41 |
+
BertTokenizer,
|
| 42 |
+
whitespace_tokenize)
|
| 43 |
+
|
| 44 |
+
if sys.version_info[0] == 2:
|
| 45 |
+
import cPickle as pickle
|
| 46 |
+
else:
|
| 47 |
+
import pickle
|
| 48 |
+
|
| 49 |
+
logger = logging.getLogger(__name__)
|
| 50 |
+
|
| 51 |
+
|
| 52 |
+
class SquadExample(object):
|
| 53 |
+
"""
|
| 54 |
+
A single training/test example for the Squad dataset.
|
| 55 |
+
For examples without an answer, the start and end position are -1.
|
| 56 |
+
"""
|
| 57 |
+
|
| 58 |
+
def __init__(self,
|
| 59 |
+
qas_id,
|
| 60 |
+
question_text,
|
| 61 |
+
doc_tokens,
|
| 62 |
+
orig_answer_text=None,
|
| 63 |
+
start_position=None,
|
| 64 |
+
end_position=None,
|
| 65 |
+
is_impossible=None):
|
| 66 |
+
self.qas_id = qas_id
|
| 67 |
+
self.question_text = question_text
|
| 68 |
+
self.doc_tokens = doc_tokens
|
| 69 |
+
self.orig_answer_text = orig_answer_text
|
| 70 |
+
self.start_position = start_position
|
| 71 |
+
self.end_position = end_position
|
| 72 |
+
self.is_impossible = is_impossible
|
| 73 |
+
|
| 74 |
+
def __str__(self):
|
| 75 |
+
return self.__repr__()
|
| 76 |
+
|
| 77 |
+
def __repr__(self):
|
| 78 |
+
s = ""
|
| 79 |
+
s += "qas_id: %s" % (self.qas_id)
|
| 80 |
+
s += ", question_text: %s" % (
|
| 81 |
+
self.question_text)
|
| 82 |
+
s += ", doc_tokens: [%s]" % (" ".join(self.doc_tokens))
|
| 83 |
+
if self.start_position:
|
| 84 |
+
s += ", start_position: %d" % (self.start_position)
|
| 85 |
+
if self.end_position:
|
| 86 |
+
s += ", end_position: %d" % (self.end_position)
|
| 87 |
+
if self.is_impossible:
|
| 88 |
+
s += ", is_impossible: %r" % (self.is_impossible)
|
| 89 |
+
return s
|
| 90 |
+
|
| 91 |
+
|
| 92 |
+
class InputFeatures(object):
|
| 93 |
+
"""A single set of features of data."""
|
| 94 |
+
|
| 95 |
+
def __init__(self,
|
| 96 |
+
unique_id,
|
| 97 |
+
example_index,
|
| 98 |
+
doc_span_index,
|
| 99 |
+
tokens,
|
| 100 |
+
token_to_orig_map,
|
| 101 |
+
token_is_max_context,
|
| 102 |
+
input_ids,
|
| 103 |
+
input_mask,
|
| 104 |
+
segment_ids,
|
| 105 |
+
start_position=None,
|
| 106 |
+
end_position=None,
|
| 107 |
+
is_impossible=None):
|
| 108 |
+
self.unique_id = unique_id
|
| 109 |
+
self.example_index = example_index
|
| 110 |
+
self.doc_span_index = doc_span_index
|
| 111 |
+
self.tokens = tokens
|
| 112 |
+
self.token_to_orig_map = token_to_orig_map
|
| 113 |
+
self.token_is_max_context = token_is_max_context
|
| 114 |
+
self.input_ids = input_ids
|
| 115 |
+
self.input_mask = input_mask
|
| 116 |
+
self.segment_ids = segment_ids
|
| 117 |
+
self.start_position = start_position
|
| 118 |
+
self.end_position = end_position
|
| 119 |
+
self.is_impossible = is_impossible
|
| 120 |
+
|
| 121 |
+
|
| 122 |
+
def read_squad_examples(input_file, is_training, version_2_with_negative):
|
| 123 |
+
"""Read a SQuAD json file into a list of SquadExample."""
|
| 124 |
+
with open(input_file, "r", encoding='utf-8') as reader:
|
| 125 |
+
input_data = json.load(reader)["data"]
|
| 126 |
+
|
| 127 |
+
def is_whitespace(c):
|
| 128 |
+
if c == " " or c == "\t" or c == "\r" or c == "\n" or ord(c) == 0x202F:
|
| 129 |
+
return True
|
| 130 |
+
return False
|
| 131 |
+
|
| 132 |
+
examples = []
|
| 133 |
+
for entry in input_data:
|
| 134 |
+
for paragraph in entry["paragraphs"]:
|
| 135 |
+
paragraph_text = paragraph["context"]
|
| 136 |
+
doc_tokens = []
|
| 137 |
+
char_to_word_offset = []
|
| 138 |
+
prev_is_whitespace = True
|
| 139 |
+
for c in paragraph_text:
|
| 140 |
+
if is_whitespace(c):
|
| 141 |
+
prev_is_whitespace = True
|
| 142 |
+
else:
|
| 143 |
+
if prev_is_whitespace:
|
| 144 |
+
doc_tokens.append(c)
|
| 145 |
+
else:
|
| 146 |
+
doc_tokens[-1] += c
|
| 147 |
+
prev_is_whitespace = False
|
| 148 |
+
char_to_word_offset.append(len(doc_tokens) - 1)
|
| 149 |
+
|
| 150 |
+
for qa in paragraph["qas"]:
|
| 151 |
+
qas_id = qa["id"]
|
| 152 |
+
question_text = qa["question"]
|
| 153 |
+
start_position = None
|
| 154 |
+
end_position = None
|
| 155 |
+
orig_answer_text = None
|
| 156 |
+
is_impossible = False
|
| 157 |
+
if is_training:
|
| 158 |
+
if version_2_with_negative:
|
| 159 |
+
is_impossible = qa["is_impossible"]
|
| 160 |
+
if (len(qa["answers"]) != 1) and (not is_impossible):
|
| 161 |
+
raise ValueError(
|
| 162 |
+
"For training, each question should have exactly 1 answer.")
|
| 163 |
+
if not is_impossible:
|
| 164 |
+
answer = qa["answers"][0]
|
| 165 |
+
orig_answer_text = answer["text"]
|
| 166 |
+
answer_offset = answer["answer_start"]
|
| 167 |
+
answer_length = len(orig_answer_text)
|
| 168 |
+
start_position = char_to_word_offset[answer_offset]
|
| 169 |
+
end_position = char_to_word_offset[answer_offset + answer_length - 1]
|
| 170 |
+
# Only add answers where the text can be exactly recovered from the
|
| 171 |
+
# document. If this CAN'T happen it's likely due to weird Unicode
|
| 172 |
+
# stuff so we will just skip the example.
|
| 173 |
+
#
|
| 174 |
+
# Note that this means for training mode, every example is NOT
|
| 175 |
+
# guaranteed to be preserved.
|
| 176 |
+
actual_text = " ".join(doc_tokens[start_position:(end_position + 1)])
|
| 177 |
+
cleaned_answer_text = " ".join(
|
| 178 |
+
whitespace_tokenize(orig_answer_text))
|
| 179 |
+
if actual_text.find(cleaned_answer_text) == -1:
|
| 180 |
+
logger.warning("Could not find answer: '%s' vs. '%s'",
|
| 181 |
+
actual_text, cleaned_answer_text)
|
| 182 |
+
continue
|
| 183 |
+
else:
|
| 184 |
+
start_position = -1
|
| 185 |
+
end_position = -1
|
| 186 |
+
orig_answer_text = ""
|
| 187 |
+
|
| 188 |
+
example = SquadExample(
|
| 189 |
+
qas_id=qas_id,
|
| 190 |
+
question_text=question_text,
|
| 191 |
+
doc_tokens=doc_tokens,
|
| 192 |
+
orig_answer_text=orig_answer_text,
|
| 193 |
+
start_position=start_position,
|
| 194 |
+
end_position=end_position,
|
| 195 |
+
is_impossible=is_impossible)
|
| 196 |
+
examples.append(example)
|
| 197 |
+
return examples
|
| 198 |
+
|
| 199 |
+
|
| 200 |
+
def convert_examples_to_features(examples, tokenizer, max_seq_length,
|
| 201 |
+
doc_stride, max_query_length, is_training):
|
| 202 |
+
"""Loads a data file into a list of `InputBatch`s."""
|
| 203 |
+
|
| 204 |
+
unique_id = 1000000000
|
| 205 |
+
|
| 206 |
+
features = []
|
| 207 |
+
for (example_index, example) in enumerate(examples):
|
| 208 |
+
query_tokens = tokenizer.tokenize(example.question_text)
|
| 209 |
+
|
| 210 |
+
if len(query_tokens) > max_query_length:
|
| 211 |
+
query_tokens = query_tokens[0:max_query_length]
|
| 212 |
+
|
| 213 |
+
tok_to_orig_index = []
|
| 214 |
+
orig_to_tok_index = []
|
| 215 |
+
all_doc_tokens = []
|
| 216 |
+
for (i, token) in enumerate(example.doc_tokens):
|
| 217 |
+
orig_to_tok_index.append(len(all_doc_tokens))
|
| 218 |
+
sub_tokens = tokenizer.tokenize(token)
|
| 219 |
+
for sub_token in sub_tokens:
|
| 220 |
+
tok_to_orig_index.append(i)
|
| 221 |
+
all_doc_tokens.append(sub_token)
|
| 222 |
+
|
| 223 |
+
tok_start_position = None
|
| 224 |
+
tok_end_position = None
|
| 225 |
+
if is_training and example.is_impossible:
|
| 226 |
+
tok_start_position = -1
|
| 227 |
+
tok_end_position = -1
|
| 228 |
+
if is_training and not example.is_impossible:
|
| 229 |
+
tok_start_position = orig_to_tok_index[example.start_position]
|
| 230 |
+
if example.end_position < len(example.doc_tokens) - 1:
|
| 231 |
+
tok_end_position = orig_to_tok_index[example.end_position + 1] - 1
|
| 232 |
+
else:
|
| 233 |
+
tok_end_position = len(all_doc_tokens) - 1
|
| 234 |
+
(tok_start_position, tok_end_position) = _improve_answer_span(
|
| 235 |
+
all_doc_tokens, tok_start_position, tok_end_position, tokenizer,
|
| 236 |
+
example.orig_answer_text)
|
| 237 |
+
|
| 238 |
+
# The -3 accounts for [CLS], [SEP] and [SEP]
|
| 239 |
+
max_tokens_for_doc = max_seq_length - len(query_tokens) - 3
|
| 240 |
+
|
| 241 |
+
# We can have documents that are longer than the maximum sequence length.
|
| 242 |
+
# To deal with this we do a sliding window approach, where we take chunks
|
| 243 |
+
# of the up to our max length with a stride of `doc_stride`.
|
| 244 |
+
_DocSpan = collections.namedtuple( # pylint: disable=invalid-name
|
| 245 |
+
"DocSpan", ["start", "length"])
|
| 246 |
+
doc_spans = []
|
| 247 |
+
start_offset = 0
|
| 248 |
+
while start_offset < len(all_doc_tokens):
|
| 249 |
+
length = len(all_doc_tokens) - start_offset
|
| 250 |
+
if length > max_tokens_for_doc:
|
| 251 |
+
length = max_tokens_for_doc
|
| 252 |
+
doc_spans.append(_DocSpan(start=start_offset, length=length))
|
| 253 |
+
if start_offset + length == len(all_doc_tokens):
|
| 254 |
+
break
|
| 255 |
+
start_offset += min(length, doc_stride)
|
| 256 |
+
|
| 257 |
+
for (doc_span_index, doc_span) in enumerate(doc_spans):
|
| 258 |
+
tokens = []
|
| 259 |
+
token_to_orig_map = {}
|
| 260 |
+
token_is_max_context = {}
|
| 261 |
+
segment_ids = []
|
| 262 |
+
tokens.append("[CLS]")
|
| 263 |
+
segment_ids.append(0)
|
| 264 |
+
for token in query_tokens:
|
| 265 |
+
tokens.append(token)
|
| 266 |
+
segment_ids.append(0)
|
| 267 |
+
tokens.append("[SEP]")
|
| 268 |
+
segment_ids.append(0)
|
| 269 |
+
|
| 270 |
+
for i in range(doc_span.length):
|
| 271 |
+
split_token_index = doc_span.start + i
|
| 272 |
+
token_to_orig_map[len(tokens)] = tok_to_orig_index[split_token_index]
|
| 273 |
+
|
| 274 |
+
is_max_context = _check_is_max_context(doc_spans, doc_span_index,
|
| 275 |
+
split_token_index)
|
| 276 |
+
token_is_max_context[len(tokens)] = is_max_context
|
| 277 |
+
tokens.append(all_doc_tokens[split_token_index])
|
| 278 |
+
segment_ids.append(1)
|
| 279 |
+
tokens.append("[SEP]")
|
| 280 |
+
segment_ids.append(1)
|
| 281 |
+
|
| 282 |
+
input_ids = tokenizer.convert_tokens_to_ids(tokens)
|
| 283 |
+
|
| 284 |
+
# The mask has 1 for real tokens and 0 for padding tokens. Only real
|
| 285 |
+
# tokens are attended to.
|
| 286 |
+
input_mask = [1] * len(input_ids)
|
| 287 |
+
|
| 288 |
+
# Zero-pad up to the sequence length.
|
| 289 |
+
while len(input_ids) < max_seq_length:
|
| 290 |
+
input_ids.append(0)
|
| 291 |
+
input_mask.append(0)
|
| 292 |
+
segment_ids.append(0)
|
| 293 |
+
|
| 294 |
+
assert len(input_ids) == max_seq_length
|
| 295 |
+
assert len(input_mask) == max_seq_length
|
| 296 |
+
assert len(segment_ids) == max_seq_length
|
| 297 |
+
|
| 298 |
+
start_position = None
|
| 299 |
+
end_position = None
|
| 300 |
+
if is_training and not example.is_impossible:
|
| 301 |
+
# For training, if our document chunk does not contain an annotation
|
| 302 |
+
# we throw it out, since there is nothing to predict.
|
| 303 |
+
doc_start = doc_span.start
|
| 304 |
+
doc_end = doc_span.start + doc_span.length - 1
|
| 305 |
+
out_of_span = False
|
| 306 |
+
if not (tok_start_position >= doc_start and
|
| 307 |
+
tok_end_position <= doc_end):
|
| 308 |
+
out_of_span = True
|
| 309 |
+
if out_of_span:
|
| 310 |
+
start_position = 0
|
| 311 |
+
end_position = 0
|
| 312 |
+
else:
|
| 313 |
+
doc_offset = len(query_tokens) + 2
|
| 314 |
+
start_position = tok_start_position - doc_start + doc_offset
|
| 315 |
+
end_position = tok_end_position - doc_start + doc_offset
|
| 316 |
+
if is_training and example.is_impossible:
|
| 317 |
+
start_position = 0
|
| 318 |
+
end_position = 0
|
| 319 |
+
if example_index < 20:
|
| 320 |
+
logger.info("*** Example ***")
|
| 321 |
+
logger.info("unique_id: %s" % (unique_id))
|
| 322 |
+
logger.info("example_index: %s" % (example_index))
|
| 323 |
+
logger.info("doc_span_index: %s" % (doc_span_index))
|
| 324 |
+
logger.info("tokens: %s" % " ".join(tokens))
|
| 325 |
+
logger.info("token_to_orig_map: %s" % " ".join([
|
| 326 |
+
"%d:%d" % (x, y) for (x, y) in token_to_orig_map.items()]))
|
| 327 |
+
logger.info("token_is_max_context: %s" % " ".join([
|
| 328 |
+
"%d:%s" % (x, y) for (x, y) in token_is_max_context.items()
|
| 329 |
+
]))
|
| 330 |
+
logger.info("input_ids: %s" % " ".join([str(x) for x in input_ids]))
|
| 331 |
+
logger.info(
|
| 332 |
+
"input_mask: %s" % " ".join([str(x) for x in input_mask]))
|
| 333 |
+
logger.info(
|
| 334 |
+
"segment_ids: %s" % " ".join([str(x) for x in segment_ids]))
|
| 335 |
+
if is_training and example.is_impossible:
|
| 336 |
+
logger.info("impossible example")
|
| 337 |
+
if is_training and not example.is_impossible:
|
| 338 |
+
answer_text = " ".join(tokens[start_position:(end_position + 1)])
|
| 339 |
+
logger.info("start_position: %d" % (start_position))
|
| 340 |
+
logger.info("end_position: %d" % (end_position))
|
| 341 |
+
logger.info(
|
| 342 |
+
"answer: %s" % (answer_text))
|
| 343 |
+
|
| 344 |
+
features.append(
|
| 345 |
+
InputFeatures(
|
| 346 |
+
unique_id=unique_id,
|
| 347 |
+
example_index=example_index,
|
| 348 |
+
doc_span_index=doc_span_index,
|
| 349 |
+
tokens=tokens,
|
| 350 |
+
token_to_orig_map=token_to_orig_map,
|
| 351 |
+
token_is_max_context=token_is_max_context,
|
| 352 |
+
input_ids=input_ids,
|
| 353 |
+
input_mask=input_mask,
|
| 354 |
+
segment_ids=segment_ids,
|
| 355 |
+
start_position=start_position,
|
| 356 |
+
end_position=end_position,
|
| 357 |
+
is_impossible=example.is_impossible))
|
| 358 |
+
unique_id += 1
|
| 359 |
+
|
| 360 |
+
return features
|
| 361 |
+
|
| 362 |
+
|
| 363 |
+
def _improve_answer_span(doc_tokens, input_start, input_end, tokenizer,
|
| 364 |
+
orig_answer_text):
|
| 365 |
+
"""Returns tokenized answer spans that better match the annotated answer."""
|
| 366 |
+
|
| 367 |
+
# The SQuAD annotations are character based. We first project them to
|
| 368 |
+
# whitespace-tokenized words. But then after WordPiece tokenization, we can
|
| 369 |
+
# often find a "better match". For example:
|
| 370 |
+
#
|
| 371 |
+
# Question: What year was John Smith born?
|
| 372 |
+
# Context: The leader was John Smith (1895-1943).
|
| 373 |
+
# Answer: 1895
|
| 374 |
+
#
|
| 375 |
+
# The original whitespace-tokenized answer will be "(1895-1943).". However
|
| 376 |
+
# after tokenization, our tokens will be "( 1895 - 1943 ) .". So we can match
|
| 377 |
+
# the exact answer, 1895.
|
| 378 |
+
#
|
| 379 |
+
# However, this is not always possible. Consider the following:
|
| 380 |
+
#
|
| 381 |
+
# Question: What country is the top exporter of electornics?
|
| 382 |
+
# Context: The Japanese electronics industry is the lagest in the world.
|
| 383 |
+
# Answer: Japan
|
| 384 |
+
#
|
| 385 |
+
# In this case, the annotator chose "Japan" as a character sub-span of
|
| 386 |
+
# the word "Japanese". Since our WordPiece tokenizer does not split
|
| 387 |
+
# "Japanese", we just use "Japanese" as the annotation. This is fairly rare
|
| 388 |
+
# in SQuAD, but does happen.
|
| 389 |
+
tok_answer_text = " ".join(tokenizer.tokenize(orig_answer_text))
|
| 390 |
+
|
| 391 |
+
for new_start in range(input_start, input_end + 1):
|
| 392 |
+
for new_end in range(input_end, new_start - 1, -1):
|
| 393 |
+
text_span = " ".join(doc_tokens[new_start:(new_end + 1)])
|
| 394 |
+
if text_span == tok_answer_text:
|
| 395 |
+
return (new_start, new_end)
|
| 396 |
+
|
| 397 |
+
return (input_start, input_end)
|
| 398 |
+
|
| 399 |
+
|
| 400 |
+
def _check_is_max_context(doc_spans, cur_span_index, position):
|
| 401 |
+
"""Check if this is the 'max context' doc span for the token."""
|
| 402 |
+
|
| 403 |
+
# Because of the sliding window approach taken to scoring documents, a single
|
| 404 |
+
# token can appear in multiple documents. E.g.
|
| 405 |
+
# Doc: the man went to the store and bought a gallon of milk
|
| 406 |
+
# Span A: the man went to the
|
| 407 |
+
# Span B: to the store and bought
|
| 408 |
+
# Span C: and bought a gallon of
|
| 409 |
+
# ...
|
| 410 |
+
#
|
| 411 |
+
# Now the word 'bought' will have two scores from spans B and C. We only
|
| 412 |
+
# want to consider the score with "maximum context", which we define as
|
| 413 |
+
# the *minimum* of its left and right context (the *sum* of left and
|
| 414 |
+
# right context will always be the same, of course).
|
| 415 |
+
#
|
| 416 |
+
# In the example the maximum context for 'bought' would be span C since
|
| 417 |
+
# it has 1 left context and 3 right context, while span B has 4 left context
|
| 418 |
+
# and 0 right context.
|
| 419 |
+
best_score = None
|
| 420 |
+
best_span_index = None
|
| 421 |
+
for (span_index, doc_span) in enumerate(doc_spans):
|
| 422 |
+
end = doc_span.start + doc_span.length - 1
|
| 423 |
+
if position < doc_span.start:
|
| 424 |
+
continue
|
| 425 |
+
if position > end:
|
| 426 |
+
continue
|
| 427 |
+
num_left_context = position - doc_span.start
|
| 428 |
+
num_right_context = end - position
|
| 429 |
+
score = min(num_left_context, num_right_context) + 0.01 * doc_span.length
|
| 430 |
+
if best_score is None or score > best_score:
|
| 431 |
+
best_score = score
|
| 432 |
+
best_span_index = span_index
|
| 433 |
+
|
| 434 |
+
return cur_span_index == best_span_index
|
| 435 |
+
|
| 436 |
+
|
| 437 |
+
RawResult = collections.namedtuple("RawResult",
|
| 438 |
+
["unique_id", "start_logits", "end_logits"])
|
| 439 |
+
|
| 440 |
+
|
| 441 |
+
def write_predictions(all_examples, all_features, all_results, n_best_size,
|
| 442 |
+
max_answer_length, do_lower_case, output_prediction_file,
|
| 443 |
+
output_nbest_file, output_null_log_odds_file, verbose_logging,
|
| 444 |
+
version_2_with_negative, null_score_diff_threshold):
|
| 445 |
+
"""Write final predictions to the json file and log-odds of null if needed."""
|
| 446 |
+
logger.info("Writing predictions to: %s" % (output_prediction_file))
|
| 447 |
+
logger.info("Writing nbest to: %s" % (output_nbest_file))
|
| 448 |
+
|
| 449 |
+
example_index_to_features = collections.defaultdict(list)
|
| 450 |
+
for feature in all_features:
|
| 451 |
+
example_index_to_features[feature.example_index].append(feature)
|
| 452 |
+
|
| 453 |
+
unique_id_to_result = {}
|
| 454 |
+
for result in all_results:
|
| 455 |
+
unique_id_to_result[result.unique_id] = result
|
| 456 |
+
|
| 457 |
+
_PrelimPrediction = collections.namedtuple( # pylint: disable=invalid-name
|
| 458 |
+
"PrelimPrediction",
|
| 459 |
+
["feature_index", "start_index", "end_index", "start_logit", "end_logit"])
|
| 460 |
+
|
| 461 |
+
all_predictions = collections.OrderedDict()
|
| 462 |
+
all_nbest_json = collections.OrderedDict()
|
| 463 |
+
scores_diff_json = collections.OrderedDict()
|
| 464 |
+
|
| 465 |
+
for (example_index, example) in enumerate(all_examples):
|
| 466 |
+
features = example_index_to_features[example_index]
|
| 467 |
+
|
| 468 |
+
prelim_predictions = []
|
| 469 |
+
# keep track of the minimum score of null start+end of position 0
|
| 470 |
+
score_null = 1000000 # large and positive
|
| 471 |
+
min_null_feature_index = 0 # the paragraph slice with min null score
|
| 472 |
+
null_start_logit = 0 # the start logit at the slice with min null score
|
| 473 |
+
null_end_logit = 0 # the end logit at the slice with min null score
|
| 474 |
+
for (feature_index, feature) in enumerate(features):
|
| 475 |
+
result = unique_id_to_result[feature.unique_id]
|
| 476 |
+
start_indexes = _get_best_indexes(result.start_logits, n_best_size)
|
| 477 |
+
end_indexes = _get_best_indexes(result.end_logits, n_best_size)
|
| 478 |
+
# if we could have irrelevant answers, get the min score of irrelevant
|
| 479 |
+
if version_2_with_negative:
|
| 480 |
+
feature_null_score = result.start_logits[0] + result.end_logits[0]
|
| 481 |
+
if feature_null_score < score_null:
|
| 482 |
+
score_null = feature_null_score
|
| 483 |
+
min_null_feature_index = feature_index
|
| 484 |
+
null_start_logit = result.start_logits[0]
|
| 485 |
+
null_end_logit = result.end_logits[0]
|
| 486 |
+
for start_index in start_indexes:
|
| 487 |
+
for end_index in end_indexes:
|
| 488 |
+
# We could hypothetically create invalid predictions, e.g., predict
|
| 489 |
+
# that the start of the span is in the question. We throw out all
|
| 490 |
+
# invalid predictions.
|
| 491 |
+
if start_index >= len(feature.tokens):
|
| 492 |
+
continue
|
| 493 |
+
if end_index >= len(feature.tokens):
|
| 494 |
+
continue
|
| 495 |
+
if start_index not in feature.token_to_orig_map:
|
| 496 |
+
continue
|
| 497 |
+
if end_index not in feature.token_to_orig_map:
|
| 498 |
+
continue
|
| 499 |
+
if not feature.token_is_max_context.get(start_index, False):
|
| 500 |
+
continue
|
| 501 |
+
if end_index < start_index:
|
| 502 |
+
continue
|
| 503 |
+
length = end_index - start_index + 1
|
| 504 |
+
if length > max_answer_length:
|
| 505 |
+
continue
|
| 506 |
+
prelim_predictions.append(
|
| 507 |
+
_PrelimPrediction(
|
| 508 |
+
feature_index=feature_index,
|
| 509 |
+
start_index=start_index,
|
| 510 |
+
end_index=end_index,
|
| 511 |
+
start_logit=result.start_logits[start_index],
|
| 512 |
+
end_logit=result.end_logits[end_index]))
|
| 513 |
+
if version_2_with_negative:
|
| 514 |
+
prelim_predictions.append(
|
| 515 |
+
_PrelimPrediction(
|
| 516 |
+
feature_index=min_null_feature_index,
|
| 517 |
+
start_index=0,
|
| 518 |
+
end_index=0,
|
| 519 |
+
start_logit=null_start_logit,
|
| 520 |
+
end_logit=null_end_logit))
|
| 521 |
+
prelim_predictions = sorted(
|
| 522 |
+
prelim_predictions,
|
| 523 |
+
key=lambda x: (x.start_logit + x.end_logit),
|
| 524 |
+
reverse=True)
|
| 525 |
+
|
| 526 |
+
_NbestPrediction = collections.namedtuple( # pylint: disable=invalid-name
|
| 527 |
+
"NbestPrediction", ["text", "start_logit", "end_logit"])
|
| 528 |
+
|
| 529 |
+
seen_predictions = {}
|
| 530 |
+
nbest = []
|
| 531 |
+
for pred in prelim_predictions:
|
| 532 |
+
if len(nbest) >= n_best_size:
|
| 533 |
+
break
|
| 534 |
+
feature = features[pred.feature_index]
|
| 535 |
+
if pred.start_index > 0: # this is a non-null prediction
|
| 536 |
+
tok_tokens = feature.tokens[pred.start_index:(pred.end_index + 1)]
|
| 537 |
+
orig_doc_start = feature.token_to_orig_map[pred.start_index]
|
| 538 |
+
orig_doc_end = feature.token_to_orig_map[pred.end_index]
|
| 539 |
+
orig_tokens = example.doc_tokens[orig_doc_start:(orig_doc_end + 1)]
|
| 540 |
+
tok_text = " ".join(tok_tokens)
|
| 541 |
+
|
| 542 |
+
# De-tokenize WordPieces that have been split off.
|
| 543 |
+
tok_text = tok_text.replace(" ##", "")
|
| 544 |
+
tok_text = tok_text.replace("##", "")
|
| 545 |
+
|
| 546 |
+
# Clean whitespace
|
| 547 |
+
tok_text = tok_text.strip()
|
| 548 |
+
tok_text = " ".join(tok_text.split())
|
| 549 |
+
orig_text = " ".join(orig_tokens)
|
| 550 |
+
|
| 551 |
+
final_text = get_final_text(tok_text, orig_text, do_lower_case, verbose_logging)
|
| 552 |
+
if final_text in seen_predictions:
|
| 553 |
+
continue
|
| 554 |
+
|
| 555 |
+
seen_predictions[final_text] = True
|
| 556 |
+
else:
|
| 557 |
+
final_text = ""
|
| 558 |
+
seen_predictions[final_text] = True
|
| 559 |
+
|
| 560 |
+
nbest.append(
|
| 561 |
+
_NbestPrediction(
|
| 562 |
+
text=final_text,
|
| 563 |
+
start_logit=pred.start_logit,
|
| 564 |
+
end_logit=pred.end_logit))
|
| 565 |
+
# if we didn't include the empty option in the n-best, include it
|
| 566 |
+
if version_2_with_negative:
|
| 567 |
+
if "" not in seen_predictions:
|
| 568 |
+
nbest.append(
|
| 569 |
+
_NbestPrediction(
|
| 570 |
+
text="",
|
| 571 |
+
start_logit=null_start_logit,
|
| 572 |
+
end_logit=null_end_logit))
|
| 573 |
+
|
| 574 |
+
# In very rare edge cases we could only have single null prediction.
|
| 575 |
+
# So we just create a nonce prediction in this case to avoid failure.
|
| 576 |
+
if len(nbest)==1:
|
| 577 |
+
nbest.insert(0,
|
| 578 |
+
_NbestPrediction(text="empty", start_logit=0.0, end_logit=0.0))
|
| 579 |
+
|
| 580 |
+
# In very rare edge cases we could have no valid predictions. So we
|
| 581 |
+
# just create a nonce prediction in this case to avoid failure.
|
| 582 |
+
if not nbest:
|
| 583 |
+
nbest.append(
|
| 584 |
+
_NbestPrediction(text="empty", start_logit=0.0, end_logit=0.0))
|
| 585 |
+
|
| 586 |
+
assert len(nbest) >= 1
|
| 587 |
+
|
| 588 |
+
total_scores = []
|
| 589 |
+
best_non_null_entry = None
|
| 590 |
+
for entry in nbest:
|
| 591 |
+
total_scores.append(entry.start_logit + entry.end_logit)
|
| 592 |
+
if not best_non_null_entry:
|
| 593 |
+
if entry.text:
|
| 594 |
+
best_non_null_entry = entry
|
| 595 |
+
|
| 596 |
+
probs = _compute_softmax(total_scores)
|
| 597 |
+
|
| 598 |
+
nbest_json = []
|
| 599 |
+
for (i, entry) in enumerate(nbest):
|
| 600 |
+
output = collections.OrderedDict()
|
| 601 |
+
output["text"] = entry.text
|
| 602 |
+
output["probability"] = probs[i]
|
| 603 |
+
output["start_logit"] = entry.start_logit
|
| 604 |
+
output["end_logit"] = entry.end_logit
|
| 605 |
+
nbest_json.append(output)
|
| 606 |
+
|
| 607 |
+
assert len(nbest_json) >= 1
|
| 608 |
+
|
| 609 |
+
if not version_2_with_negative:
|
| 610 |
+
all_predictions[example.qas_id] = nbest_json[0]["text"]
|
| 611 |
+
else:
|
| 612 |
+
# predict "" iff the null score - the score of best non-null > threshold
|
| 613 |
+
score_diff = score_null - best_non_null_entry.start_logit - (
|
| 614 |
+
best_non_null_entry.end_logit)
|
| 615 |
+
scores_diff_json[example.qas_id] = score_diff
|
| 616 |
+
if score_diff > null_score_diff_threshold:
|
| 617 |
+
all_predictions[example.qas_id] = ""
|
| 618 |
+
else:
|
| 619 |
+
all_predictions[example.qas_id] = best_non_null_entry.text
|
| 620 |
+
all_nbest_json[example.qas_id] = nbest_json
|
| 621 |
+
|
| 622 |
+
with open(output_prediction_file, "w") as writer:
|
| 623 |
+
writer.write(json.dumps(all_predictions, indent=4) + "\n")
|
| 624 |
+
|
| 625 |
+
with open(output_nbest_file, "w") as writer:
|
| 626 |
+
writer.write(json.dumps(all_nbest_json, indent=4) + "\n")
|
| 627 |
+
|
| 628 |
+
if version_2_with_negative:
|
| 629 |
+
with open(output_null_log_odds_file, "w") as writer:
|
| 630 |
+
writer.write(json.dumps(scores_diff_json, indent=4) + "\n")
|
| 631 |
+
|
| 632 |
+
|
| 633 |
+
def get_final_text(pred_text, orig_text, do_lower_case, verbose_logging=False):
|
| 634 |
+
"""Project the tokenized prediction back to the original text."""
|
| 635 |
+
|
| 636 |
+
# When we created the data, we kept track of the alignment between original
|
| 637 |
+
# (whitespace tokenized) tokens and our WordPiece tokenized tokens. So
|
| 638 |
+
# now `orig_text` contains the span of our original text corresponding to the
|
| 639 |
+
# span that we predicted.
|
| 640 |
+
#
|
| 641 |
+
# However, `orig_text` may contain extra characters that we don't want in
|
| 642 |
+
# our prediction.
|
| 643 |
+
#
|
| 644 |
+
# For example, let's say:
|
| 645 |
+
# pred_text = steve smith
|
| 646 |
+
# orig_text = Steve Smith's
|
| 647 |
+
#
|
| 648 |
+
# We don't want to return `orig_text` because it contains the extra "'s".
|
| 649 |
+
#
|
| 650 |
+
# We don't want to return `pred_text` because it's already been normalized
|
| 651 |
+
# (the SQuAD eval script also does punctuation stripping/lower casing but
|
| 652 |
+
# our tokenizer does additional normalization like stripping accent
|
| 653 |
+
# characters).
|
| 654 |
+
#
|
| 655 |
+
# What we really want to return is "Steve Smith".
|
| 656 |
+
#
|
| 657 |
+
# Therefore, we have to apply a semi-complicated alignment heuristic between
|
| 658 |
+
# `pred_text` and `orig_text` to get a character-to-character alignment. This
|
| 659 |
+
# can fail in certain cases in which case we just return `orig_text`.
|
| 660 |
+
|
| 661 |
+
def _strip_spaces(text):
|
| 662 |
+
ns_chars = []
|
| 663 |
+
ns_to_s_map = collections.OrderedDict()
|
| 664 |
+
for (i, c) in enumerate(text):
|
| 665 |
+
if c == " ":
|
| 666 |
+
continue
|
| 667 |
+
ns_to_s_map[len(ns_chars)] = i
|
| 668 |
+
ns_chars.append(c)
|
| 669 |
+
ns_text = "".join(ns_chars)
|
| 670 |
+
return (ns_text, ns_to_s_map)
|
| 671 |
+
|
| 672 |
+
# We first tokenize `orig_text`, strip whitespace from the result
|
| 673 |
+
# and `pred_text`, and check if they are the same length. If they are
|
| 674 |
+
# NOT the same length, the heuristic has failed. If they are the same
|
| 675 |
+
# length, we assume the characters are one-to-one aligned.
|
| 676 |
+
tokenizer = BasicTokenizer(do_lower_case=do_lower_case)
|
| 677 |
+
|
| 678 |
+
tok_text = " ".join(tokenizer.tokenize(orig_text))
|
| 679 |
+
|
| 680 |
+
start_position = tok_text.find(pred_text)
|
| 681 |
+
if start_position == -1:
|
| 682 |
+
if verbose_logging:
|
| 683 |
+
logger.info(
|
| 684 |
+
"Unable to find text: '%s' in '%s'" % (pred_text, orig_text))
|
| 685 |
+
return orig_text
|
| 686 |
+
end_position = start_position + len(pred_text) - 1
|
| 687 |
+
|
| 688 |
+
(orig_ns_text, orig_ns_to_s_map) = _strip_spaces(orig_text)
|
| 689 |
+
(tok_ns_text, tok_ns_to_s_map) = _strip_spaces(tok_text)
|
| 690 |
+
|
| 691 |
+
if len(orig_ns_text) != len(tok_ns_text):
|
| 692 |
+
if verbose_logging:
|
| 693 |
+
logger.info("Length not equal after stripping spaces: '%s' vs '%s'",
|
| 694 |
+
orig_ns_text, tok_ns_text)
|
| 695 |
+
return orig_text
|
| 696 |
+
|
| 697 |
+
# We then project the characters in `pred_text` back to `orig_text` using
|
| 698 |
+
# the character-to-character alignment.
|
| 699 |
+
tok_s_to_ns_map = {}
|
| 700 |
+
for (i, tok_index) in tok_ns_to_s_map.items():
|
| 701 |
+
tok_s_to_ns_map[tok_index] = i
|
| 702 |
+
|
| 703 |
+
orig_start_position = None
|
| 704 |
+
if start_position in tok_s_to_ns_map:
|
| 705 |
+
ns_start_position = tok_s_to_ns_map[start_position]
|
| 706 |
+
if ns_start_position in orig_ns_to_s_map:
|
| 707 |
+
orig_start_position = orig_ns_to_s_map[ns_start_position]
|
| 708 |
+
|
| 709 |
+
if orig_start_position is None:
|
| 710 |
+
if verbose_logging:
|
| 711 |
+
logger.info("Couldn't map start position")
|
| 712 |
+
return orig_text
|
| 713 |
+
|
| 714 |
+
orig_end_position = None
|
| 715 |
+
if end_position in tok_s_to_ns_map:
|
| 716 |
+
ns_end_position = tok_s_to_ns_map[end_position]
|
| 717 |
+
if ns_end_position in orig_ns_to_s_map:
|
| 718 |
+
orig_end_position = orig_ns_to_s_map[ns_end_position]
|
| 719 |
+
|
| 720 |
+
if orig_end_position is None:
|
| 721 |
+
if verbose_logging:
|
| 722 |
+
logger.info("Couldn't map end position")
|
| 723 |
+
return orig_text
|
| 724 |
+
|
| 725 |
+
output_text = orig_text[orig_start_position:(orig_end_position + 1)]
|
| 726 |
+
return output_text
|
| 727 |
+
|
| 728 |
+
|
| 729 |
+
def _get_best_indexes(logits, n_best_size):
|
| 730 |
+
"""Get the n-best logits from a list."""
|
| 731 |
+
index_and_score = sorted(enumerate(logits), key=lambda x: x[1], reverse=True)
|
| 732 |
+
|
| 733 |
+
best_indexes = []
|
| 734 |
+
for i in range(len(index_and_score)):
|
| 735 |
+
if i >= n_best_size:
|
| 736 |
+
break
|
| 737 |
+
best_indexes.append(index_and_score[i][0])
|
| 738 |
+
return best_indexes
|
| 739 |
+
|
| 740 |
+
|
| 741 |
+
def _compute_softmax(scores):
|
| 742 |
+
"""Compute softmax probability over raw logits."""
|
| 743 |
+
if not scores:
|
| 744 |
+
return []
|
| 745 |
+
|
| 746 |
+
max_score = None
|
| 747 |
+
for score in scores:
|
| 748 |
+
if max_score is None or score > max_score:
|
| 749 |
+
max_score = score
|
| 750 |
+
|
| 751 |
+
exp_scores = []
|
| 752 |
+
total_sum = 0.0
|
| 753 |
+
for score in scores:
|
| 754 |
+
x = math.exp(score - max_score)
|
| 755 |
+
exp_scores.append(x)
|
| 756 |
+
total_sum += x
|
| 757 |
+
|
| 758 |
+
probs = []
|
| 759 |
+
for score in exp_scores:
|
| 760 |
+
probs.append(score / total_sum)
|
| 761 |
+
return probs
|
| 762 |
+
|
| 763 |
+
def main():
|
| 764 |
+
parser = argparse.ArgumentParser()
|
| 765 |
+
|
| 766 |
+
## Required parameters
|
| 767 |
+
parser.add_argument("--bert_model", default=None, type=str, required=True,
|
| 768 |
+
help="Bert pre-trained model selected in the list: bert-base-uncased, "
|
| 769 |
+
"bert-large-uncased, bert-base-cased, bert-large-cased, bert-base-multilingual-uncased, "
|
| 770 |
+
"bert-base-multilingual-cased, bert-base-chinese.")
|
| 771 |
+
parser.add_argument("--output_dir", default=None, type=str, required=True,
|
| 772 |
+
help="The output directory where the model checkpoints and predictions will be written.")
|
| 773 |
+
|
| 774 |
+
## Other parameters
|
| 775 |
+
parser.add_argument("--train_file", default=None, type=str, help="SQuAD json for training. E.g., train-v1.1.json")
|
| 776 |
+
parser.add_argument("--predict_file", default=None, type=str,
|
| 777 |
+
help="SQuAD json for predictions. E.g., dev-v1.1.json or test-v1.1.json")
|
| 778 |
+
parser.add_argument("--max_seq_length", default=384, type=int,
|
| 779 |
+
help="The maximum total input sequence length after WordPiece tokenization. Sequences "
|
| 780 |
+
"longer than this will be truncated, and sequences shorter than this will be padded.")
|
| 781 |
+
parser.add_argument("--doc_stride", default=128, type=int,
|
| 782 |
+
help="When splitting up a long document into chunks, how much stride to take between chunks.")
|
| 783 |
+
parser.add_argument("--max_query_length", default=64, type=int,
|
| 784 |
+
help="The maximum number of tokens for the question. Questions longer than this will "
|
| 785 |
+
"be truncated to this length.")
|
| 786 |
+
parser.add_argument("--do_train", action='store_true', help="Whether to run training.")
|
| 787 |
+
parser.add_argument("--do_predict", action='store_true', help="Whether to run eval on the dev set.")
|
| 788 |
+
parser.add_argument("--train_batch_size", default=32, type=int, help="Total batch size for training.")
|
| 789 |
+
parser.add_argument("--predict_batch_size", default=8, type=int, help="Total batch size for predictions.")
|
| 790 |
+
parser.add_argument("--learning_rate", default=5e-5, type=float, help="The initial learning rate for Adam.")
|
| 791 |
+
parser.add_argument("--num_train_epochs", default=3.0, type=float,
|
| 792 |
+
help="Total number of training epochs to perform.")
|
| 793 |
+
parser.add_argument("--warmup_proportion", default=0.1, type=float,
|
| 794 |
+
help="Proportion of training to perform linear learning rate warmup for. E.g., 0.1 = 10%% "
|
| 795 |
+
"of training.")
|
| 796 |
+
parser.add_argument("--n_best_size", default=20, type=int,
|
| 797 |
+
help="The total number of n-best predictions to generate in the nbest_predictions.json "
|
| 798 |
+
"output file.")
|
| 799 |
+
parser.add_argument("--max_answer_length", default=30, type=int,
|
| 800 |
+
help="The maximum length of an answer that can be generated. This is needed because the start "
|
| 801 |
+
"and end predictions are not conditioned on one another.")
|
| 802 |
+
parser.add_argument("--verbose_logging", action='store_true',
|
| 803 |
+
help="If true, all of the warnings related to data processing will be printed. "
|
| 804 |
+
"A number of warnings are expected for a normal SQuAD evaluation.")
|
| 805 |
+
parser.add_argument("--no_cuda",
|
| 806 |
+
action='store_true',
|
| 807 |
+
help="Whether not to use CUDA when available")
|
| 808 |
+
parser.add_argument('--seed',
|
| 809 |
+
type=int,
|
| 810 |
+
default=42,
|
| 811 |
+
help="random seed for initialization")
|
| 812 |
+
parser.add_argument('--gradient_accumulation_steps',
|
| 813 |
+
type=int,
|
| 814 |
+
default=1,
|
| 815 |
+
help="Number of updates steps to accumulate before performing a backward/update pass.")
|
| 816 |
+
parser.add_argument("--do_lower_case",
|
| 817 |
+
action='store_true',
|
| 818 |
+
help="Whether to lower case the input text. True for uncased models, False for cased models.")
|
| 819 |
+
parser.add_argument("--local_rank",
|
| 820 |
+
type=int,
|
| 821 |
+
default=-1,
|
| 822 |
+
help="local_rank for distributed training on gpus")
|
| 823 |
+
parser.add_argument('--fp16',
|
| 824 |
+
action='store_true',
|
| 825 |
+
help="Whether to use 16-bit float precision instead of 32-bit")
|
| 826 |
+
parser.add_argument('--loss_scale',
|
| 827 |
+
type=float, default=0,
|
| 828 |
+
help="Loss scaling to improve fp16 numeric stability. Only used when fp16 set to True.\n"
|
| 829 |
+
"0 (default value): dynamic loss scaling.\n"
|
| 830 |
+
"Positive power of 2: static loss scaling value.\n")
|
| 831 |
+
parser.add_argument('--version_2_with_negative',
|
| 832 |
+
action='store_true',
|
| 833 |
+
help='If true, the SQuAD examples contain some that do not have an answer.')
|
| 834 |
+
parser.add_argument('--null_score_diff_threshold',
|
| 835 |
+
type=float, default=0.0,
|
| 836 |
+
help="If null_score - best_non_null is greater than the threshold predict null.")
|
| 837 |
+
parser.add_argument('--server_ip', type=str, default='', help="Can be used for distant debugging.")
|
| 838 |
+
parser.add_argument('--server_port', type=str, default='', help="Can be used for distant debugging.")
|
| 839 |
+
args = parser.parse_args()
|
| 840 |
+
print(args)
|
| 841 |
+
|
| 842 |
+
if args.server_ip and args.server_port:
|
| 843 |
+
# Distant debugging - see https://code.visualstudio.com/docs/python/debugging#_attach-to-a-local-script
|
| 844 |
+
import ptvsd
|
| 845 |
+
print("Waiting for debugger attach")
|
| 846 |
+
ptvsd.enable_attach(address=(args.server_ip, args.server_port), redirect_output=True)
|
| 847 |
+
ptvsd.wait_for_attach()
|
| 848 |
+
|
| 849 |
+
if args.local_rank == -1 or args.no_cuda:
|
| 850 |
+
device = torch.device("cuda" if torch.cuda.is_available() and not args.no_cuda else "cpu")
|
| 851 |
+
n_gpu = torch.cuda.device_count()
|
| 852 |
+
else:
|
| 853 |
+
torch.cuda.set_device(args.local_rank)
|
| 854 |
+
device = torch.device("cuda", args.local_rank)
|
| 855 |
+
n_gpu = 1
|
| 856 |
+
# Initializes the distributed backend which will take care of sychronizing nodes/GPUs
|
| 857 |
+
torch.distributed.init_process_group(backend='nccl')
|
| 858 |
+
|
| 859 |
+
logging.basicConfig(format = '%(asctime)s - %(levelname)s - %(name)s - %(message)s',
|
| 860 |
+
datefmt = '%m/%d/%Y %H:%M:%S',
|
| 861 |
+
level = logging.INFO if args.local_rank in [-1, 0] else logging.WARN)
|
| 862 |
+
|
| 863 |
+
logger.info("device: {} n_gpu: {}, distributed training: {}, 16-bits training: {}".format(
|
| 864 |
+
device, n_gpu, bool(args.local_rank != -1), args.fp16))
|
| 865 |
+
|
| 866 |
+
if args.gradient_accumulation_steps < 1:
|
| 867 |
+
raise ValueError("Invalid gradient_accumulation_steps parameter: {}, should be >= 1".format(
|
| 868 |
+
args.gradient_accumulation_steps))
|
| 869 |
+
|
| 870 |
+
args.train_batch_size = args.train_batch_size // args.gradient_accumulation_steps
|
| 871 |
+
|
| 872 |
+
random.seed(args.seed)
|
| 873 |
+
np.random.seed(args.seed)
|
| 874 |
+
torch.manual_seed(args.seed)
|
| 875 |
+
if n_gpu > 0:
|
| 876 |
+
torch.cuda.manual_seed_all(args.seed)
|
| 877 |
+
|
| 878 |
+
if not args.do_train and not args.do_predict:
|
| 879 |
+
raise ValueError("At least one of `do_train` or `do_predict` must be True.")
|
| 880 |
+
|
| 881 |
+
if args.do_train:
|
| 882 |
+
if not args.train_file:
|
| 883 |
+
raise ValueError(
|
| 884 |
+
"If `do_train` is True, then `train_file` must be specified.")
|
| 885 |
+
if args.do_predict:
|
| 886 |
+
if not args.predict_file:
|
| 887 |
+
raise ValueError(
|
| 888 |
+
"If `do_predict` is True, then `predict_file` must be specified.")
|
| 889 |
+
|
| 890 |
+
if os.path.exists(args.output_dir) and os.listdir(args.output_dir) and args.do_train:
|
| 891 |
+
raise ValueError("Output directory () already exists and is not empty.")
|
| 892 |
+
if not os.path.exists(args.output_dir):
|
| 893 |
+
os.makedirs(args.output_dir)
|
| 894 |
+
|
| 895 |
+
tokenizer = BertTokenizer.from_pretrained(args.bert_model, do_lower_case=args.do_lower_case)
|
| 896 |
+
|
| 897 |
+
train_examples = None
|
| 898 |
+
num_train_optimization_steps = None
|
| 899 |
+
if args.do_train:
|
| 900 |
+
train_examples = read_squad_examples(
|
| 901 |
+
input_file=args.train_file, is_training=True, version_2_with_negative=args.version_2_with_negative)
|
| 902 |
+
num_train_optimization_steps = int(
|
| 903 |
+
len(train_examples) / args.train_batch_size / args.gradient_accumulation_steps) * args.num_train_epochs
|
| 904 |
+
if args.local_rank != -1:
|
| 905 |
+
num_train_optimization_steps = num_train_optimization_steps // torch.distributed.get_world_size()
|
| 906 |
+
|
| 907 |
+
# Prepare model
|
| 908 |
+
model = BertForQuestionAnswering.from_pretrained(args.bert_model,
|
| 909 |
+
cache_dir=os.path.join(str(PYTORCH_PRETRAINED_BERT_CACHE), 'distributed_{}'.format(args.local_rank)))
|
| 910 |
+
|
| 911 |
+
if args.fp16:
|
| 912 |
+
model.half()
|
| 913 |
+
model.to(device)
|
| 914 |
+
if args.local_rank != -1:
|
| 915 |
+
try:
|
| 916 |
+
from apex.parallel import DistributedDataParallel as DDP
|
| 917 |
+
except ImportError:
|
| 918 |
+
raise ImportError("Please install apex from https://www.github.com/nvidia/apex to use distributed and fp16 training.")
|
| 919 |
+
|
| 920 |
+
model = DDP(model)
|
| 921 |
+
elif n_gpu > 1:
|
| 922 |
+
model = torch.nn.DataParallel(model)
|
| 923 |
+
|
| 924 |
+
# Prepare optimizer
|
| 925 |
+
param_optimizer = list(model.named_parameters())
|
| 926 |
+
|
| 927 |
+
# hack to remove pooler, which is not used
|
| 928 |
+
# thus it produce None grad that break apex
|
| 929 |
+
param_optimizer = [n for n in param_optimizer if 'pooler' not in n[0]]
|
| 930 |
+
|
| 931 |
+
no_decay = ['bias', 'LayerNorm.bias', 'LayerNorm.weight']
|
| 932 |
+
optimizer_grouped_parameters = [
|
| 933 |
+
{'params': [p for n, p in param_optimizer if not any(nd in n for nd in no_decay)], 'weight_decay': 0.01},
|
| 934 |
+
{'params': [p for n, p in param_optimizer if any(nd in n for nd in no_decay)], 'weight_decay': 0.0}
|
| 935 |
+
]
|
| 936 |
+
|
| 937 |
+
if args.fp16:
|
| 938 |
+
try:
|
| 939 |
+
from apex.optimizers import FP16_Optimizer
|
| 940 |
+
from apex.optimizers import FusedAdam
|
| 941 |
+
except ImportError:
|
| 942 |
+
raise ImportError("Please install apex from https://www.github.com/nvidia/apex to use distributed and fp16 training.")
|
| 943 |
+
|
| 944 |
+
optimizer = FusedAdam(optimizer_grouped_parameters,
|
| 945 |
+
lr=args.learning_rate,
|
| 946 |
+
bias_correction=False,
|
| 947 |
+
max_grad_norm=1.0)
|
| 948 |
+
if args.loss_scale == 0:
|
| 949 |
+
optimizer = FP16_Optimizer(optimizer, dynamic_loss_scale=True)
|
| 950 |
+
else:
|
| 951 |
+
optimizer = FP16_Optimizer(optimizer, static_loss_scale=args.loss_scale)
|
| 952 |
+
else:
|
| 953 |
+
optimizer = BertAdam(optimizer_grouped_parameters,
|
| 954 |
+
lr=args.learning_rate,
|
| 955 |
+
warmup=args.warmup_proportion,
|
| 956 |
+
t_total=num_train_optimization_steps)
|
| 957 |
+
|
| 958 |
+
global_step = 0
|
| 959 |
+
if args.do_train:
|
| 960 |
+
cached_train_features_file = args.train_file+'_{0}_{1}_{2}_{3}'.format(
|
| 961 |
+
list(filter(None, args.bert_model.split('/'))).pop(), str(args.max_seq_length), str(args.doc_stride), str(args.max_query_length))
|
| 962 |
+
train_features = None
|
| 963 |
+
try:
|
| 964 |
+
with open(cached_train_features_file, "rb") as reader:
|
| 965 |
+
train_features = pickle.load(reader)
|
| 966 |
+
except:
|
| 967 |
+
train_features = convert_examples_to_features(
|
| 968 |
+
examples=train_examples,
|
| 969 |
+
tokenizer=tokenizer,
|
| 970 |
+
max_seq_length=args.max_seq_length,
|
| 971 |
+
doc_stride=args.doc_stride,
|
| 972 |
+
max_query_length=args.max_query_length,
|
| 973 |
+
is_training=True)
|
| 974 |
+
if args.local_rank == -1 or torch.distributed.get_rank() == 0:
|
| 975 |
+
logger.info(" Saving train features into cached file %s", cached_train_features_file)
|
| 976 |
+
with open(cached_train_features_file, "wb") as writer:
|
| 977 |
+
pickle.dump(train_features, writer)
|
| 978 |
+
logger.info("***** Running training *****")
|
| 979 |
+
logger.info(" Num orig examples = %d", len(train_examples))
|
| 980 |
+
logger.info(" Num split examples = %d", len(train_features))
|
| 981 |
+
logger.info(" Batch size = %d", args.train_batch_size)
|
| 982 |
+
logger.info(" Num steps = %d", num_train_optimization_steps)
|
| 983 |
+
all_input_ids = torch.tensor([f.input_ids for f in train_features], dtype=torch.long)
|
| 984 |
+
all_input_mask = torch.tensor([f.input_mask for f in train_features], dtype=torch.long)
|
| 985 |
+
all_segment_ids = torch.tensor([f.segment_ids for f in train_features], dtype=torch.long)
|
| 986 |
+
all_start_positions = torch.tensor([f.start_position for f in train_features], dtype=torch.long)
|
| 987 |
+
all_end_positions = torch.tensor([f.end_position for f in train_features], dtype=torch.long)
|
| 988 |
+
train_data = TensorDataset(all_input_ids, all_input_mask, all_segment_ids,
|
| 989 |
+
all_start_positions, all_end_positions)
|
| 990 |
+
if args.local_rank == -1:
|
| 991 |
+
train_sampler = RandomSampler(train_data)
|
| 992 |
+
else:
|
| 993 |
+
train_sampler = DistributedSampler(train_data)
|
| 994 |
+
train_dataloader = DataLoader(train_data, sampler=train_sampler, batch_size=args.train_batch_size)
|
| 995 |
+
|
| 996 |
+
model.train()
|
| 997 |
+
for _ in trange(int(args.num_train_epochs), desc="Epoch"):
|
| 998 |
+
for step, batch in enumerate(tqdm(train_dataloader, desc="Iteration", disable=args.local_rank not in [-1, 0])):
|
| 999 |
+
if n_gpu == 1:
|
| 1000 |
+
batch = tuple(t.to(device) for t in batch) # multi-gpu does scattering it-self
|
| 1001 |
+
input_ids, input_mask, segment_ids, start_positions, end_positions = batch
|
| 1002 |
+
loss = model(input_ids, segment_ids, input_mask, start_positions, end_positions)
|
| 1003 |
+
if n_gpu > 1:
|
| 1004 |
+
loss = loss.mean() # mean() to average on multi-gpu.
|
| 1005 |
+
if args.gradient_accumulation_steps > 1:
|
| 1006 |
+
loss = loss / args.gradient_accumulation_steps
|
| 1007 |
+
|
| 1008 |
+
if args.fp16:
|
| 1009 |
+
optimizer.backward(loss)
|
| 1010 |
+
else:
|
| 1011 |
+
loss.backward()
|
| 1012 |
+
if (step + 1) % args.gradient_accumulation_steps == 0:
|
| 1013 |
+
if args.fp16:
|
| 1014 |
+
# modify learning rate with special warm up BERT uses
|
| 1015 |
+
# if args.fp16 is False, BertAdam is used and handles this automatically
|
| 1016 |
+
lr_this_step = args.learning_rate * warmup_linear(global_step/num_train_optimization_steps, args.warmup_proportion)
|
| 1017 |
+
for param_group in optimizer.param_groups:
|
| 1018 |
+
param_group['lr'] = lr_this_step
|
| 1019 |
+
optimizer.step()
|
| 1020 |
+
optimizer.zero_grad()
|
| 1021 |
+
global_step += 1
|
| 1022 |
+
|
| 1023 |
+
if args.do_train and (args.local_rank == -1 or torch.distributed.get_rank() == 0):
|
| 1024 |
+
# Save a trained model, configuration and tokenizer
|
| 1025 |
+
model_to_save = model.module if hasattr(model, 'module') else model # Only save the model it-self
|
| 1026 |
+
|
| 1027 |
+
# If we save using the predefined names, we can load using `from_pretrained`
|
| 1028 |
+
output_model_file = os.path.join(args.output_dir, WEIGHTS_NAME)
|
| 1029 |
+
output_config_file = os.path.join(args.output_dir, CONFIG_NAME)
|
| 1030 |
+
|
| 1031 |
+
torch.save(model_to_save.state_dict(), output_model_file)
|
| 1032 |
+
model_to_save.config.to_json_file(output_config_file)
|
| 1033 |
+
tokenizer.save_vocabulary(args.output_dir)
|
| 1034 |
+
|
| 1035 |
+
# Load a trained model and vocabulary that you have fine-tuned
|
| 1036 |
+
model = BertForQuestionAnswering.from_pretrained(args.output_dir)
|
| 1037 |
+
tokenizer = BertTokenizer.from_pretrained(args.output_dir, do_lower_case=args.do_lower_case)
|
| 1038 |
+
else:
|
| 1039 |
+
model = BertForQuestionAnswering.from_pretrained(args.bert_model)
|
| 1040 |
+
|
| 1041 |
+
model.to(device)
|
| 1042 |
+
|
| 1043 |
+
if args.do_predict and (args.local_rank == -1 or torch.distributed.get_rank() == 0):
|
| 1044 |
+
eval_examples = read_squad_examples(
|
| 1045 |
+
input_file=args.predict_file, is_training=False, version_2_with_negative=args.version_2_with_negative)
|
| 1046 |
+
eval_features = convert_examples_to_features(
|
| 1047 |
+
examples=eval_examples,
|
| 1048 |
+
tokenizer=tokenizer,
|
| 1049 |
+
max_seq_length=args.max_seq_length,
|
| 1050 |
+
doc_stride=args.doc_stride,
|
| 1051 |
+
max_query_length=args.max_query_length,
|
| 1052 |
+
is_training=False)
|
| 1053 |
+
|
| 1054 |
+
logger.info("***** Running predictions *****")
|
| 1055 |
+
logger.info(" Num orig examples = %d", len(eval_examples))
|
| 1056 |
+
logger.info(" Num split examples = %d", len(eval_features))
|
| 1057 |
+
logger.info(" Batch size = %d", args.predict_batch_size)
|
| 1058 |
+
|
| 1059 |
+
all_input_ids = torch.tensor([f.input_ids for f in eval_features], dtype=torch.long)
|
| 1060 |
+
all_input_mask = torch.tensor([f.input_mask for f in eval_features], dtype=torch.long)
|
| 1061 |
+
all_segment_ids = torch.tensor([f.segment_ids for f in eval_features], dtype=torch.long)
|
| 1062 |
+
all_example_index = torch.arange(all_input_ids.size(0), dtype=torch.long)
|
| 1063 |
+
eval_data = TensorDataset(all_input_ids, all_input_mask, all_segment_ids, all_example_index)
|
| 1064 |
+
# Run prediction for full data
|
| 1065 |
+
eval_sampler = SequentialSampler(eval_data)
|
| 1066 |
+
eval_dataloader = DataLoader(eval_data, sampler=eval_sampler, batch_size=args.predict_batch_size)
|
| 1067 |
+
|
| 1068 |
+
model.eval()
|
| 1069 |
+
all_results = []
|
| 1070 |
+
logger.info("Start evaluating")
|
| 1071 |
+
for input_ids, input_mask, segment_ids, example_indices in tqdm(eval_dataloader, desc="Evaluating", disable=args.local_rank not in [-1, 0]):
|
| 1072 |
+
if len(all_results) % 1000 == 0:
|
| 1073 |
+
logger.info("Processing example: %d" % (len(all_results)))
|
| 1074 |
+
input_ids = input_ids.to(device)
|
| 1075 |
+
input_mask = input_mask.to(device)
|
| 1076 |
+
segment_ids = segment_ids.to(device)
|
| 1077 |
+
with torch.no_grad():
|
| 1078 |
+
batch_start_logits, batch_end_logits = model(input_ids, segment_ids, input_mask)
|
| 1079 |
+
for i, example_index in enumerate(example_indices):
|
| 1080 |
+
start_logits = batch_start_logits[i].detach().cpu().tolist()
|
| 1081 |
+
end_logits = batch_end_logits[i].detach().cpu().tolist()
|
| 1082 |
+
eval_feature = eval_features[example_index.item()]
|
| 1083 |
+
unique_id = int(eval_feature.unique_id)
|
| 1084 |
+
all_results.append(RawResult(unique_id=unique_id,
|
| 1085 |
+
start_logits=start_logits,
|
| 1086 |
+
end_logits=end_logits))
|
| 1087 |
+
output_prediction_file = os.path.join(args.output_dir, "predictions.json")
|
| 1088 |
+
output_nbest_file = os.path.join(args.output_dir, "nbest_predictions.json")
|
| 1089 |
+
output_null_log_odds_file = os.path.join(args.output_dir, "null_odds.json")
|
| 1090 |
+
write_predictions(eval_examples, eval_features, all_results,
|
| 1091 |
+
args.n_best_size, args.max_answer_length,
|
| 1092 |
+
args.do_lower_case, output_prediction_file,
|
| 1093 |
+
output_nbest_file, output_null_log_odds_file, args.verbose_logging,
|
| 1094 |
+
args.version_2_with_negative, args.null_score_diff_threshold)
|
| 1095 |
+
|
| 1096 |
+
|
| 1097 |
+
if __name__ == "__main__":
|
| 1098 |
+
main()
|
dark-secrets-of-BERT-master/dark-secrets-of-BERT-master/examples/run_swag.py
ADDED
|
@@ -0,0 +1,551 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# coding=utf-8
|
| 2 |
+
# Copyright 2018 The Google AI Language Team Authors and The HuggingFace Inc. team.
|
| 3 |
+
# Copyright (c) 2018, NVIDIA CORPORATION. All rights reserved.
|
| 4 |
+
#
|
| 5 |
+
# Licensed under the Apache License, Version 2.0 (the "License");
|
| 6 |
+
# you may not use this file except in compliance with the License.
|
| 7 |
+
# You may obtain a copy of the License at
|
| 8 |
+
#
|
| 9 |
+
# http://www.apache.org/licenses/LICENSE-2.0
|
| 10 |
+
#
|
| 11 |
+
# Unless required by applicable law or agreed to in writing, software
|
| 12 |
+
# distributed under the License is distributed on an "AS IS" BASIS,
|
| 13 |
+
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
| 14 |
+
# See the License for the specific language governing permissions and
|
| 15 |
+
# limitations under the License.
|
| 16 |
+
"""BERT finetuning runner."""
|
| 17 |
+
|
| 18 |
+
from __future__ import absolute_import
|
| 19 |
+
|
| 20 |
+
import argparse
|
| 21 |
+
import csv
|
| 22 |
+
import logging
|
| 23 |
+
import os
|
| 24 |
+
import random
|
| 25 |
+
import sys
|
| 26 |
+
from io import open
|
| 27 |
+
|
| 28 |
+
import numpy as np
|
| 29 |
+
import torch
|
| 30 |
+
from torch.utils.data import (DataLoader, RandomSampler, SequentialSampler,
|
| 31 |
+
TensorDataset)
|
| 32 |
+
from torch.utils.data.distributed import DistributedSampler
|
| 33 |
+
from tqdm import tqdm, trange
|
| 34 |
+
|
| 35 |
+
from pytorch_pretrained_bert.file_utils import PYTORCH_PRETRAINED_BERT_CACHE, WEIGHTS_NAME, CONFIG_NAME
|
| 36 |
+
from pytorch_pretrained_bert.modeling import BertForMultipleChoice, BertConfig
|
| 37 |
+
from pytorch_pretrained_bert.optimization import BertAdam, warmup_linear
|
| 38 |
+
from pytorch_pretrained_bert.tokenization import BertTokenizer
|
| 39 |
+
|
| 40 |
+
logging.basicConfig(format = '%(asctime)s - %(levelname)s - %(name)s - %(message)s',
|
| 41 |
+
datefmt = '%m/%d/%Y %H:%M:%S',
|
| 42 |
+
level = logging.INFO)
|
| 43 |
+
logger = logging.getLogger(__name__)
|
| 44 |
+
|
| 45 |
+
|
| 46 |
+
class SwagExample(object):
|
| 47 |
+
"""A single training/test example for the SWAG dataset."""
|
| 48 |
+
def __init__(self,
|
| 49 |
+
swag_id,
|
| 50 |
+
context_sentence,
|
| 51 |
+
start_ending,
|
| 52 |
+
ending_0,
|
| 53 |
+
ending_1,
|
| 54 |
+
ending_2,
|
| 55 |
+
ending_3,
|
| 56 |
+
label = None):
|
| 57 |
+
self.swag_id = swag_id
|
| 58 |
+
self.context_sentence = context_sentence
|
| 59 |
+
self.start_ending = start_ending
|
| 60 |
+
self.endings = [
|
| 61 |
+
ending_0,
|
| 62 |
+
ending_1,
|
| 63 |
+
ending_2,
|
| 64 |
+
ending_3,
|
| 65 |
+
]
|
| 66 |
+
self.label = label
|
| 67 |
+
|
| 68 |
+
def __str__(self):
|
| 69 |
+
return self.__repr__()
|
| 70 |
+
|
| 71 |
+
def __repr__(self):
|
| 72 |
+
l = [
|
| 73 |
+
"swag_id: {}".format(self.swag_id),
|
| 74 |
+
"context_sentence: {}".format(self.context_sentence),
|
| 75 |
+
"start_ending: {}".format(self.start_ending),
|
| 76 |
+
"ending_0: {}".format(self.endings[0]),
|
| 77 |
+
"ending_1: {}".format(self.endings[1]),
|
| 78 |
+
"ending_2: {}".format(self.endings[2]),
|
| 79 |
+
"ending_3: {}".format(self.endings[3]),
|
| 80 |
+
]
|
| 81 |
+
|
| 82 |
+
if self.label is not None:
|
| 83 |
+
l.append("label: {}".format(self.label))
|
| 84 |
+
|
| 85 |
+
return ", ".join(l)
|
| 86 |
+
|
| 87 |
+
|
| 88 |
+
class InputFeatures(object):
|
| 89 |
+
def __init__(self,
|
| 90 |
+
example_id,
|
| 91 |
+
choices_features,
|
| 92 |
+
label
|
| 93 |
+
|
| 94 |
+
):
|
| 95 |
+
self.example_id = example_id
|
| 96 |
+
self.choices_features = [
|
| 97 |
+
{
|
| 98 |
+
'input_ids': input_ids,
|
| 99 |
+
'input_mask': input_mask,
|
| 100 |
+
'segment_ids': segment_ids
|
| 101 |
+
}
|
| 102 |
+
for _, input_ids, input_mask, segment_ids in choices_features
|
| 103 |
+
]
|
| 104 |
+
self.label = label
|
| 105 |
+
|
| 106 |
+
|
| 107 |
+
def read_swag_examples(input_file, is_training):
|
| 108 |
+
with open(input_file, 'r', encoding='utf-8') as f:
|
| 109 |
+
reader = csv.reader(f)
|
| 110 |
+
lines = []
|
| 111 |
+
for line in reader:
|
| 112 |
+
if sys.version_info[0] == 2:
|
| 113 |
+
line = list(unicode(cell, 'utf-8') for cell in line)
|
| 114 |
+
lines.append(line)
|
| 115 |
+
|
| 116 |
+
if is_training and lines[0][-1] != 'label':
|
| 117 |
+
raise ValueError(
|
| 118 |
+
"For training, the input file must contain a label column."
|
| 119 |
+
)
|
| 120 |
+
|
| 121 |
+
examples = [
|
| 122 |
+
SwagExample(
|
| 123 |
+
swag_id = line[2],
|
| 124 |
+
context_sentence = line[4],
|
| 125 |
+
start_ending = line[5], # in the swag dataset, the
|
| 126 |
+
# common beginning of each
|
| 127 |
+
# choice is stored in "sent2".
|
| 128 |
+
ending_0 = line[7],
|
| 129 |
+
ending_1 = line[8],
|
| 130 |
+
ending_2 = line[9],
|
| 131 |
+
ending_3 = line[10],
|
| 132 |
+
label = int(line[11]) if is_training else None
|
| 133 |
+
) for line in lines[1:] # we skip the line with the column names
|
| 134 |
+
]
|
| 135 |
+
|
| 136 |
+
return examples
|
| 137 |
+
|
| 138 |
+
def convert_examples_to_features(examples, tokenizer, max_seq_length,
|
| 139 |
+
is_training):
|
| 140 |
+
"""Loads a data file into a list of `InputBatch`s."""
|
| 141 |
+
|
| 142 |
+
# Swag is a multiple choice task. To perform this task using Bert,
|
| 143 |
+
# we will use the formatting proposed in "Improving Language
|
| 144 |
+
# Understanding by Generative Pre-Training" and suggested by
|
| 145 |
+
# @jacobdevlin-google in this issue
|
| 146 |
+
# https://github.com/google-research/bert/issues/38.
|
| 147 |
+
#
|
| 148 |
+
# Each choice will correspond to a sample on which we run the
|
| 149 |
+
# inference. For a given Swag example, we will create the 4
|
| 150 |
+
# following inputs:
|
| 151 |
+
# - [CLS] context [SEP] choice_1 [SEP]
|
| 152 |
+
# - [CLS] context [SEP] choice_2 [SEP]
|
| 153 |
+
# - [CLS] context [SEP] choice_3 [SEP]
|
| 154 |
+
# - [CLS] context [SEP] choice_4 [SEP]
|
| 155 |
+
# The model will output a single value for each input. To get the
|
| 156 |
+
# final decision of the model, we will run a softmax over these 4
|
| 157 |
+
# outputs.
|
| 158 |
+
features = []
|
| 159 |
+
for example_index, example in enumerate(examples):
|
| 160 |
+
context_tokens = tokenizer.tokenize(example.context_sentence)
|
| 161 |
+
start_ending_tokens = tokenizer.tokenize(example.start_ending)
|
| 162 |
+
|
| 163 |
+
choices_features = []
|
| 164 |
+
for ending_index, ending in enumerate(example.endings):
|
| 165 |
+
# We create a copy of the context tokens in order to be
|
| 166 |
+
# able to shrink it according to ending_tokens
|
| 167 |
+
context_tokens_choice = context_tokens[:]
|
| 168 |
+
ending_tokens = start_ending_tokens + tokenizer.tokenize(ending)
|
| 169 |
+
# Modifies `context_tokens_choice` and `ending_tokens` in
|
| 170 |
+
# place so that the total length is less than the
|
| 171 |
+
# specified length. Account for [CLS], [SEP], [SEP] with
|
| 172 |
+
# "- 3"
|
| 173 |
+
_truncate_seq_pair(context_tokens_choice, ending_tokens, max_seq_length - 3)
|
| 174 |
+
|
| 175 |
+
tokens = ["[CLS]"] + context_tokens_choice + ["[SEP]"] + ending_tokens + ["[SEP]"]
|
| 176 |
+
segment_ids = [0] * (len(context_tokens_choice) + 2) + [1] * (len(ending_tokens) + 1)
|
| 177 |
+
|
| 178 |
+
input_ids = tokenizer.convert_tokens_to_ids(tokens)
|
| 179 |
+
input_mask = [1] * len(input_ids)
|
| 180 |
+
|
| 181 |
+
# Zero-pad up to the sequence length.
|
| 182 |
+
padding = [0] * (max_seq_length - len(input_ids))
|
| 183 |
+
input_ids += padding
|
| 184 |
+
input_mask += padding
|
| 185 |
+
segment_ids += padding
|
| 186 |
+
|
| 187 |
+
assert len(input_ids) == max_seq_length
|
| 188 |
+
assert len(input_mask) == max_seq_length
|
| 189 |
+
assert len(segment_ids) == max_seq_length
|
| 190 |
+
|
| 191 |
+
choices_features.append((tokens, input_ids, input_mask, segment_ids))
|
| 192 |
+
|
| 193 |
+
label = example.label
|
| 194 |
+
if example_index < 5:
|
| 195 |
+
logger.info("*** Example ***")
|
| 196 |
+
logger.info("swag_id: {}".format(example.swag_id))
|
| 197 |
+
for choice_idx, (tokens, input_ids, input_mask, segment_ids) in enumerate(choices_features):
|
| 198 |
+
logger.info("choice: {}".format(choice_idx))
|
| 199 |
+
logger.info("tokens: {}".format(' '.join(tokens)))
|
| 200 |
+
logger.info("input_ids: {}".format(' '.join(map(str, input_ids))))
|
| 201 |
+
logger.info("input_mask: {}".format(' '.join(map(str, input_mask))))
|
| 202 |
+
logger.info("segment_ids: {}".format(' '.join(map(str, segment_ids))))
|
| 203 |
+
if is_training:
|
| 204 |
+
logger.info("label: {}".format(label))
|
| 205 |
+
|
| 206 |
+
features.append(
|
| 207 |
+
InputFeatures(
|
| 208 |
+
example_id = example.swag_id,
|
| 209 |
+
choices_features = choices_features,
|
| 210 |
+
label = label
|
| 211 |
+
)
|
| 212 |
+
)
|
| 213 |
+
|
| 214 |
+
return features
|
| 215 |
+
|
| 216 |
+
def _truncate_seq_pair(tokens_a, tokens_b, max_length):
|
| 217 |
+
"""Truncates a sequence pair in place to the maximum length."""
|
| 218 |
+
|
| 219 |
+
# This is a simple heuristic which will always truncate the longer sequence
|
| 220 |
+
# one token at a time. This makes more sense than truncating an equal percent
|
| 221 |
+
# of tokens from each, since if one sequence is very short then each token
|
| 222 |
+
# that's truncated likely contains more information than a longer sequence.
|
| 223 |
+
while True:
|
| 224 |
+
total_length = len(tokens_a) + len(tokens_b)
|
| 225 |
+
if total_length <= max_length:
|
| 226 |
+
break
|
| 227 |
+
if len(tokens_a) > len(tokens_b):
|
| 228 |
+
tokens_a.pop()
|
| 229 |
+
else:
|
| 230 |
+
tokens_b.pop()
|
| 231 |
+
|
| 232 |
+
def accuracy(out, labels):
|
| 233 |
+
outputs = np.argmax(out, axis=1)
|
| 234 |
+
return np.sum(outputs == labels)
|
| 235 |
+
|
| 236 |
+
def select_field(features, field):
|
| 237 |
+
return [
|
| 238 |
+
[
|
| 239 |
+
choice[field]
|
| 240 |
+
for choice in feature.choices_features
|
| 241 |
+
]
|
| 242 |
+
for feature in features
|
| 243 |
+
]
|
| 244 |
+
|
| 245 |
+
def main():
|
| 246 |
+
parser = argparse.ArgumentParser()
|
| 247 |
+
|
| 248 |
+
## Required parameters
|
| 249 |
+
parser.add_argument("--data_dir",
|
| 250 |
+
default=None,
|
| 251 |
+
type=str,
|
| 252 |
+
required=True,
|
| 253 |
+
help="The input data dir. Should contain the .csv files (or other data files) for the task.")
|
| 254 |
+
parser.add_argument("--bert_model", default=None, type=str, required=True,
|
| 255 |
+
help="Bert pre-trained model selected in the list: bert-base-uncased, "
|
| 256 |
+
"bert-large-uncased, bert-base-cased, bert-large-cased, bert-base-multilingual-uncased, "
|
| 257 |
+
"bert-base-multilingual-cased, bert-base-chinese.")
|
| 258 |
+
parser.add_argument("--output_dir",
|
| 259 |
+
default=None,
|
| 260 |
+
type=str,
|
| 261 |
+
required=True,
|
| 262 |
+
help="The output directory where the model checkpoints will be written.")
|
| 263 |
+
|
| 264 |
+
## Other parameters
|
| 265 |
+
parser.add_argument("--max_seq_length",
|
| 266 |
+
default=128,
|
| 267 |
+
type=int,
|
| 268 |
+
help="The maximum total input sequence length after WordPiece tokenization. \n"
|
| 269 |
+
"Sequences longer than this will be truncated, and sequences shorter \n"
|
| 270 |
+
"than this will be padded.")
|
| 271 |
+
parser.add_argument("--do_train",
|
| 272 |
+
action='store_true',
|
| 273 |
+
help="Whether to run training.")
|
| 274 |
+
parser.add_argument("--do_eval",
|
| 275 |
+
action='store_true',
|
| 276 |
+
help="Whether to run eval on the dev set.")
|
| 277 |
+
parser.add_argument("--do_lower_case",
|
| 278 |
+
action='store_true',
|
| 279 |
+
help="Set this flag if you are using an uncased model.")
|
| 280 |
+
parser.add_argument("--train_batch_size",
|
| 281 |
+
default=32,
|
| 282 |
+
type=int,
|
| 283 |
+
help="Total batch size for training.")
|
| 284 |
+
parser.add_argument("--eval_batch_size",
|
| 285 |
+
default=8,
|
| 286 |
+
type=int,
|
| 287 |
+
help="Total batch size for eval.")
|
| 288 |
+
parser.add_argument("--learning_rate",
|
| 289 |
+
default=5e-5,
|
| 290 |
+
type=float,
|
| 291 |
+
help="The initial learning rate for Adam.")
|
| 292 |
+
parser.add_argument("--num_train_epochs",
|
| 293 |
+
default=3.0,
|
| 294 |
+
type=float,
|
| 295 |
+
help="Total number of training epochs to perform.")
|
| 296 |
+
parser.add_argument("--warmup_proportion",
|
| 297 |
+
default=0.1,
|
| 298 |
+
type=float,
|
| 299 |
+
help="Proportion of training to perform linear learning rate warmup for. "
|
| 300 |
+
"E.g., 0.1 = 10%% of training.")
|
| 301 |
+
parser.add_argument("--no_cuda",
|
| 302 |
+
action='store_true',
|
| 303 |
+
help="Whether not to use CUDA when available")
|
| 304 |
+
parser.add_argument("--local_rank",
|
| 305 |
+
type=int,
|
| 306 |
+
default=-1,
|
| 307 |
+
help="local_rank for distributed training on gpus")
|
| 308 |
+
parser.add_argument('--seed',
|
| 309 |
+
type=int,
|
| 310 |
+
default=42,
|
| 311 |
+
help="random seed for initialization")
|
| 312 |
+
parser.add_argument('--gradient_accumulation_steps',
|
| 313 |
+
type=int,
|
| 314 |
+
default=1,
|
| 315 |
+
help="Number of updates steps to accumulate before performing a backward/update pass.")
|
| 316 |
+
parser.add_argument('--fp16',
|
| 317 |
+
action='store_true',
|
| 318 |
+
help="Whether to use 16-bit float precision instead of 32-bit")
|
| 319 |
+
parser.add_argument('--loss_scale',
|
| 320 |
+
type=float, default=0,
|
| 321 |
+
help="Loss scaling to improve fp16 numeric stability. Only used when fp16 set to True.\n"
|
| 322 |
+
"0 (default value): dynamic loss scaling.\n"
|
| 323 |
+
"Positive power of 2: static loss scaling value.\n")
|
| 324 |
+
|
| 325 |
+
args = parser.parse_args()
|
| 326 |
+
|
| 327 |
+
if args.local_rank == -1 or args.no_cuda:
|
| 328 |
+
device = torch.device("cuda" if torch.cuda.is_available() and not args.no_cuda else "cpu")
|
| 329 |
+
n_gpu = torch.cuda.device_count()
|
| 330 |
+
else:
|
| 331 |
+
torch.cuda.set_device(args.local_rank)
|
| 332 |
+
device = torch.device("cuda", args.local_rank)
|
| 333 |
+
n_gpu = 1
|
| 334 |
+
# Initializes the distributed backend which will take care of sychronizing nodes/GPUs
|
| 335 |
+
torch.distributed.init_process_group(backend='nccl')
|
| 336 |
+
logger.info("device: {} n_gpu: {}, distributed training: {}, 16-bits training: {}".format(
|
| 337 |
+
device, n_gpu, bool(args.local_rank != -1), args.fp16))
|
| 338 |
+
|
| 339 |
+
if args.gradient_accumulation_steps < 1:
|
| 340 |
+
raise ValueError("Invalid gradient_accumulation_steps parameter: {}, should be >= 1".format(
|
| 341 |
+
args.gradient_accumulation_steps))
|
| 342 |
+
|
| 343 |
+
args.train_batch_size = args.train_batch_size // args.gradient_accumulation_steps
|
| 344 |
+
|
| 345 |
+
random.seed(args.seed)
|
| 346 |
+
np.random.seed(args.seed)
|
| 347 |
+
torch.manual_seed(args.seed)
|
| 348 |
+
if n_gpu > 0:
|
| 349 |
+
torch.cuda.manual_seed_all(args.seed)
|
| 350 |
+
|
| 351 |
+
if not args.do_train and not args.do_eval:
|
| 352 |
+
raise ValueError("At least one of `do_train` or `do_eval` must be True.")
|
| 353 |
+
|
| 354 |
+
if os.path.exists(args.output_dir) and os.listdir(args.output_dir):
|
| 355 |
+
raise ValueError("Output directory ({}) already exists and is not empty.".format(args.output_dir))
|
| 356 |
+
if not os.path.exists(args.output_dir):
|
| 357 |
+
os.makedirs(args.output_dir)
|
| 358 |
+
|
| 359 |
+
tokenizer = BertTokenizer.from_pretrained(args.bert_model, do_lower_case=args.do_lower_case)
|
| 360 |
+
|
| 361 |
+
train_examples = None
|
| 362 |
+
num_train_optimization_steps = None
|
| 363 |
+
if args.do_train:
|
| 364 |
+
train_examples = read_swag_examples(os.path.join(args.data_dir, 'train.csv'), is_training = True)
|
| 365 |
+
num_train_optimization_steps = int(
|
| 366 |
+
len(train_examples) / args.train_batch_size / args.gradient_accumulation_steps) * args.num_train_epochs
|
| 367 |
+
if args.local_rank != -1:
|
| 368 |
+
num_train_optimization_steps = num_train_optimization_steps // torch.distributed.get_world_size()
|
| 369 |
+
|
| 370 |
+
# Prepare model
|
| 371 |
+
model = BertForMultipleChoice.from_pretrained(args.bert_model,
|
| 372 |
+
cache_dir=os.path.join(str(PYTORCH_PRETRAINED_BERT_CACHE), 'distributed_{}'.format(args.local_rank)),
|
| 373 |
+
num_choices=4)
|
| 374 |
+
if args.fp16:
|
| 375 |
+
model.half()
|
| 376 |
+
model.to(device)
|
| 377 |
+
if args.local_rank != -1:
|
| 378 |
+
try:
|
| 379 |
+
from apex.parallel import DistributedDataParallel as DDP
|
| 380 |
+
except ImportError:
|
| 381 |
+
raise ImportError("Please install apex from https://www.github.com/nvidia/apex to use distributed and fp16 training.")
|
| 382 |
+
|
| 383 |
+
model = DDP(model)
|
| 384 |
+
elif n_gpu > 1:
|
| 385 |
+
model = torch.nn.DataParallel(model)
|
| 386 |
+
|
| 387 |
+
# Prepare optimizer
|
| 388 |
+
param_optimizer = list(model.named_parameters())
|
| 389 |
+
|
| 390 |
+
# hack to remove pooler, which is not used
|
| 391 |
+
# thus it produce None grad that break apex
|
| 392 |
+
param_optimizer = [n for n in param_optimizer if 'pooler' not in n[0]]
|
| 393 |
+
|
| 394 |
+
no_decay = ['bias', 'LayerNorm.bias', 'LayerNorm.weight']
|
| 395 |
+
optimizer_grouped_parameters = [
|
| 396 |
+
{'params': [p for n, p in param_optimizer if not any(nd in n for nd in no_decay)], 'weight_decay': 0.01},
|
| 397 |
+
{'params': [p for n, p in param_optimizer if any(nd in n for nd in no_decay)], 'weight_decay': 0.0}
|
| 398 |
+
]
|
| 399 |
+
if args.fp16:
|
| 400 |
+
try:
|
| 401 |
+
from apex.optimizers import FP16_Optimizer
|
| 402 |
+
from apex.optimizers import FusedAdam
|
| 403 |
+
except ImportError:
|
| 404 |
+
raise ImportError("Please install apex from https://www.github.com/nvidia/apex to use distributed and fp16 training.")
|
| 405 |
+
|
| 406 |
+
optimizer = FusedAdam(optimizer_grouped_parameters,
|
| 407 |
+
lr=args.learning_rate,
|
| 408 |
+
bias_correction=False,
|
| 409 |
+
max_grad_norm=1.0)
|
| 410 |
+
if args.loss_scale == 0:
|
| 411 |
+
optimizer = FP16_Optimizer(optimizer, dynamic_loss_scale=True)
|
| 412 |
+
else:
|
| 413 |
+
optimizer = FP16_Optimizer(optimizer, static_loss_scale=args.loss_scale)
|
| 414 |
+
else:
|
| 415 |
+
optimizer = BertAdam(optimizer_grouped_parameters,
|
| 416 |
+
lr=args.learning_rate,
|
| 417 |
+
warmup=args.warmup_proportion,
|
| 418 |
+
t_total=num_train_optimization_steps)
|
| 419 |
+
|
| 420 |
+
global_step = 0
|
| 421 |
+
if args.do_train:
|
| 422 |
+
train_features = convert_examples_to_features(
|
| 423 |
+
train_examples, tokenizer, args.max_seq_length, True)
|
| 424 |
+
logger.info("***** Running training *****")
|
| 425 |
+
logger.info(" Num examples = %d", len(train_examples))
|
| 426 |
+
logger.info(" Batch size = %d", args.train_batch_size)
|
| 427 |
+
logger.info(" Num steps = %d", num_train_optimization_steps)
|
| 428 |
+
all_input_ids = torch.tensor(select_field(train_features, 'input_ids'), dtype=torch.long)
|
| 429 |
+
all_input_mask = torch.tensor(select_field(train_features, 'input_mask'), dtype=torch.long)
|
| 430 |
+
all_segment_ids = torch.tensor(select_field(train_features, 'segment_ids'), dtype=torch.long)
|
| 431 |
+
all_label = torch.tensor([f.label for f in train_features], dtype=torch.long)
|
| 432 |
+
train_data = TensorDataset(all_input_ids, all_input_mask, all_segment_ids, all_label)
|
| 433 |
+
if args.local_rank == -1:
|
| 434 |
+
train_sampler = RandomSampler(train_data)
|
| 435 |
+
else:
|
| 436 |
+
train_sampler = DistributedSampler(train_data)
|
| 437 |
+
train_dataloader = DataLoader(train_data, sampler=train_sampler, batch_size=args.train_batch_size)
|
| 438 |
+
|
| 439 |
+
model.train()
|
| 440 |
+
for _ in trange(int(args.num_train_epochs), desc="Epoch"):
|
| 441 |
+
tr_loss = 0
|
| 442 |
+
nb_tr_examples, nb_tr_steps = 0, 0
|
| 443 |
+
for step, batch in enumerate(tqdm(train_dataloader, desc="Iteration")):
|
| 444 |
+
batch = tuple(t.to(device) for t in batch)
|
| 445 |
+
input_ids, input_mask, segment_ids, label_ids = batch
|
| 446 |
+
loss = model(input_ids, segment_ids, input_mask, label_ids)
|
| 447 |
+
if n_gpu > 1:
|
| 448 |
+
loss = loss.mean() # mean() to average on multi-gpu.
|
| 449 |
+
if args.fp16 and args.loss_scale != 1.0:
|
| 450 |
+
# rescale loss for fp16 training
|
| 451 |
+
# see https://docs.nvidia.com/deeplearning/sdk/mixed-precision-training/index.html
|
| 452 |
+
loss = loss * args.loss_scale
|
| 453 |
+
if args.gradient_accumulation_steps > 1:
|
| 454 |
+
loss = loss / args.gradient_accumulation_steps
|
| 455 |
+
tr_loss += loss.item()
|
| 456 |
+
nb_tr_examples += input_ids.size(0)
|
| 457 |
+
nb_tr_steps += 1
|
| 458 |
+
|
| 459 |
+
if args.fp16:
|
| 460 |
+
optimizer.backward(loss)
|
| 461 |
+
else:
|
| 462 |
+
loss.backward()
|
| 463 |
+
if (step + 1) % args.gradient_accumulation_steps == 0:
|
| 464 |
+
if args.fp16:
|
| 465 |
+
# modify learning rate with special warm up BERT uses
|
| 466 |
+
# if args.fp16 is False, BertAdam is used that handles this automatically
|
| 467 |
+
lr_this_step = args.learning_rate * warmup_linear(global_step/num_train_optimization_steps, args.warmup_proportion)
|
| 468 |
+
for param_group in optimizer.param_groups:
|
| 469 |
+
param_group['lr'] = lr_this_step
|
| 470 |
+
optimizer.step()
|
| 471 |
+
optimizer.zero_grad()
|
| 472 |
+
global_step += 1
|
| 473 |
+
|
| 474 |
+
|
| 475 |
+
if args.do_train:
|
| 476 |
+
# Save a trained model, configuration and tokenizer
|
| 477 |
+
model_to_save = model.module if hasattr(model, 'module') else model # Only save the model it-self
|
| 478 |
+
|
| 479 |
+
# If we save using the predefined names, we can load using `from_pretrained`
|
| 480 |
+
output_model_file = os.path.join(args.output_dir, WEIGHTS_NAME)
|
| 481 |
+
output_config_file = os.path.join(args.output_dir, CONFIG_NAME)
|
| 482 |
+
|
| 483 |
+
torch.save(model_to_save.state_dict(), output_model_file)
|
| 484 |
+
model_to_save.config.to_json_file(output_config_file)
|
| 485 |
+
tokenizer.save_vocabulary(args.output_dir)
|
| 486 |
+
|
| 487 |
+
# Load a trained model and vocabulary that you have fine-tuned
|
| 488 |
+
model = BertForMultipleChoice.from_pretrained(args.output_dir, num_choices=4)
|
| 489 |
+
tokenizer = BertTokenizer.from_pretrained(args.output_dir, do_lower_case=args.do_lower_case)
|
| 490 |
+
else:
|
| 491 |
+
model = BertForMultipleChoice.from_pretrained(args.bert_model, num_choices=4)
|
| 492 |
+
model.to(device)
|
| 493 |
+
|
| 494 |
+
|
| 495 |
+
if args.do_eval and (args.local_rank == -1 or torch.distributed.get_rank() == 0):
|
| 496 |
+
eval_examples = read_swag_examples(os.path.join(args.data_dir, 'val.csv'), is_training = True)
|
| 497 |
+
eval_features = convert_examples_to_features(
|
| 498 |
+
eval_examples, tokenizer, args.max_seq_length, True)
|
| 499 |
+
logger.info("***** Running evaluation *****")
|
| 500 |
+
logger.info(" Num examples = %d", len(eval_examples))
|
| 501 |
+
logger.info(" Batch size = %d", args.eval_batch_size)
|
| 502 |
+
all_input_ids = torch.tensor(select_field(eval_features, 'input_ids'), dtype=torch.long)
|
| 503 |
+
all_input_mask = torch.tensor(select_field(eval_features, 'input_mask'), dtype=torch.long)
|
| 504 |
+
all_segment_ids = torch.tensor(select_field(eval_features, 'segment_ids'), dtype=torch.long)
|
| 505 |
+
all_label = torch.tensor([f.label for f in eval_features], dtype=torch.long)
|
| 506 |
+
eval_data = TensorDataset(all_input_ids, all_input_mask, all_segment_ids, all_label)
|
| 507 |
+
# Run prediction for full data
|
| 508 |
+
eval_sampler = SequentialSampler(eval_data)
|
| 509 |
+
eval_dataloader = DataLoader(eval_data, sampler=eval_sampler, batch_size=args.eval_batch_size)
|
| 510 |
+
|
| 511 |
+
model.eval()
|
| 512 |
+
eval_loss, eval_accuracy = 0, 0
|
| 513 |
+
nb_eval_steps, nb_eval_examples = 0, 0
|
| 514 |
+
for input_ids, input_mask, segment_ids, label_ids in tqdm(eval_dataloader, desc="Evaluating"):
|
| 515 |
+
input_ids = input_ids.to(device)
|
| 516 |
+
input_mask = input_mask.to(device)
|
| 517 |
+
segment_ids = segment_ids.to(device)
|
| 518 |
+
label_ids = label_ids.to(device)
|
| 519 |
+
|
| 520 |
+
with torch.no_grad():
|
| 521 |
+
tmp_eval_loss = model(input_ids, segment_ids, input_mask, label_ids)
|
| 522 |
+
logits = model(input_ids, segment_ids, input_mask)
|
| 523 |
+
|
| 524 |
+
logits = logits.detach().cpu().numpy()
|
| 525 |
+
label_ids = label_ids.to('cpu').numpy()
|
| 526 |
+
tmp_eval_accuracy = accuracy(logits, label_ids)
|
| 527 |
+
|
| 528 |
+
eval_loss += tmp_eval_loss.mean().item()
|
| 529 |
+
eval_accuracy += tmp_eval_accuracy
|
| 530 |
+
|
| 531 |
+
nb_eval_examples += input_ids.size(0)
|
| 532 |
+
nb_eval_steps += 1
|
| 533 |
+
|
| 534 |
+
eval_loss = eval_loss / nb_eval_steps
|
| 535 |
+
eval_accuracy = eval_accuracy / nb_eval_examples
|
| 536 |
+
|
| 537 |
+
result = {'eval_loss': eval_loss,
|
| 538 |
+
'eval_accuracy': eval_accuracy,
|
| 539 |
+
'global_step': global_step,
|
| 540 |
+
'loss': tr_loss/nb_tr_steps}
|
| 541 |
+
|
| 542 |
+
output_eval_file = os.path.join(args.output_dir, "eval_results.txt")
|
| 543 |
+
with open(output_eval_file, "w") as writer:
|
| 544 |
+
logger.info("***** Eval results *****")
|
| 545 |
+
for key in sorted(result.keys()):
|
| 546 |
+
logger.info(" %s = %s", key, str(result[key]))
|
| 547 |
+
writer.write("%s = %s\n" % (key, str(result[key])))
|
| 548 |
+
|
| 549 |
+
|
| 550 |
+
if __name__ == "__main__":
|
| 551 |
+
main()
|
dark-secrets-of-BERT-master/dark-secrets-of-BERT-master/examples/run_transfo_xl.py
ADDED
|
@@ -0,0 +1,153 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# coding=utf-8
|
| 2 |
+
# Copyright 2018 Google AI, Google Brain and Carnegie Mellon University Authors and the HuggingFace Inc. team.
|
| 3 |
+
# Copyright (c) 2018, NVIDIA CORPORATION. All rights reserved.
|
| 4 |
+
#
|
| 5 |
+
# Licensed under the Apache License, Version 2.0 (the "License");
|
| 6 |
+
# you may not use this file except in compliance with the License.
|
| 7 |
+
# You may obtain a copy of the License at
|
| 8 |
+
#
|
| 9 |
+
# http://www.apache.org/licenses/LICENSE-2.0
|
| 10 |
+
#
|
| 11 |
+
# Unless required by applicable law or agreed to in writing, software
|
| 12 |
+
# distributed under the License is distributed on an "AS IS" BASIS,
|
| 13 |
+
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
| 14 |
+
# See the License for the specific language governing permissions and
|
| 15 |
+
# limitations under the License.
|
| 16 |
+
""" PyTorch Transformer XL model evaluation script.
|
| 17 |
+
Adapted from https://github.com/kimiyoung/transformer-xl.
|
| 18 |
+
In particular https://github.com/kimiyoung/transformer-xl/blob/master/pytorch/eval.py
|
| 19 |
+
|
| 20 |
+
This script with default values evaluates a pretrained Transformer-XL on WikiText 103
|
| 21 |
+
"""
|
| 22 |
+
from __future__ import absolute_import, division, print_function, unicode_literals
|
| 23 |
+
|
| 24 |
+
import argparse
|
| 25 |
+
import logging
|
| 26 |
+
import time
|
| 27 |
+
import math
|
| 28 |
+
|
| 29 |
+
import torch
|
| 30 |
+
|
| 31 |
+
from pytorch_pretrained_bert import TransfoXLLMHeadModel, TransfoXLCorpus, TransfoXLTokenizer
|
| 32 |
+
|
| 33 |
+
logging.basicConfig(format = '%(asctime)s - %(levelname)s - %(name)s - %(message)s',
|
| 34 |
+
datefmt = '%m/%d/%Y %H:%M:%S',
|
| 35 |
+
level = logging.INFO)
|
| 36 |
+
logger = logging.getLogger(__name__)
|
| 37 |
+
|
| 38 |
+
def main():
|
| 39 |
+
parser = argparse.ArgumentParser(description='PyTorch Transformer Language Model')
|
| 40 |
+
parser.add_argument('--model_name', type=str, default='transfo-xl-wt103',
|
| 41 |
+
help='pretrained model name')
|
| 42 |
+
parser.add_argument('--split', type=str, default='test',
|
| 43 |
+
choices=['all', 'valid', 'test'],
|
| 44 |
+
help='which split to evaluate')
|
| 45 |
+
parser.add_argument('--batch_size', type=int, default=10,
|
| 46 |
+
help='batch size')
|
| 47 |
+
parser.add_argument('--tgt_len', type=int, default=128,
|
| 48 |
+
help='number of tokens to predict')
|
| 49 |
+
parser.add_argument('--ext_len', type=int, default=0,
|
| 50 |
+
help='length of the extended context')
|
| 51 |
+
parser.add_argument('--mem_len', type=int, default=1600,
|
| 52 |
+
help='length of the retained previous heads')
|
| 53 |
+
parser.add_argument('--clamp_len', type=int, default=1000,
|
| 54 |
+
help='max positional embedding index')
|
| 55 |
+
parser.add_argument('--no_cuda', action='store_true',
|
| 56 |
+
help='Do not use CUDA even though CUA is available')
|
| 57 |
+
parser.add_argument('--work_dir', type=str, required=True,
|
| 58 |
+
help='path to the work_dir')
|
| 59 |
+
parser.add_argument('--no_log', action='store_true',
|
| 60 |
+
help='do not log the eval result')
|
| 61 |
+
parser.add_argument('--same_length', action='store_true',
|
| 62 |
+
help='set same length attention with masking')
|
| 63 |
+
parser.add_argument('--server_ip', type=str, default='', help="Can be used for distant debugging.")
|
| 64 |
+
parser.add_argument('--server_port', type=str, default='', help="Can be used for distant debugging.")
|
| 65 |
+
args = parser.parse_args()
|
| 66 |
+
assert args.ext_len >= 0, 'extended context length must be non-negative'
|
| 67 |
+
|
| 68 |
+
if args.server_ip and args.server_port:
|
| 69 |
+
# Distant debugging - see https://code.visualstudio.com/docs/python/debugging#_attach-to-a-local-script
|
| 70 |
+
import ptvsd
|
| 71 |
+
print("Waiting for debugger attach")
|
| 72 |
+
ptvsd.enable_attach(address=(args.server_ip, args.server_port), redirect_output=True)
|
| 73 |
+
ptvsd.wait_for_attach()
|
| 74 |
+
|
| 75 |
+
device = torch.device("cuda" if torch.cuda.is_available() and not args.no_cuda else "cpu")
|
| 76 |
+
logger.info("device: {}".format(device))
|
| 77 |
+
|
| 78 |
+
# Load a pre-processed dataset
|
| 79 |
+
# You can also build the corpus yourself using TransfoXLCorpus methods
|
| 80 |
+
# The pre-processing involve computing word frequencies to prepare the Adaptive input and SoftMax
|
| 81 |
+
# and tokenizing the dataset
|
| 82 |
+
# The pre-processed corpus is a convertion (using the conversion script )
|
| 83 |
+
tokenizer = TransfoXLTokenizer.from_pretrained(args.model_name)
|
| 84 |
+
corpus = TransfoXLCorpus.from_pretrained(args.model_name)
|
| 85 |
+
ntokens = len(corpus.vocab)
|
| 86 |
+
|
| 87 |
+
va_iter = corpus.get_iterator('valid', args.batch_size, args.tgt_len,
|
| 88 |
+
device=device, ext_len=args.ext_len)
|
| 89 |
+
te_iter = corpus.get_iterator('test', args.batch_size, args.tgt_len,
|
| 90 |
+
device=device, ext_len=args.ext_len)
|
| 91 |
+
|
| 92 |
+
# Load a pre-trained model
|
| 93 |
+
model = TransfoXLLMHeadModel.from_pretrained(args.model_name)
|
| 94 |
+
model = model.to(device)
|
| 95 |
+
|
| 96 |
+
logger.info('Evaluating with bsz {} tgt_len {} ext_len {} mem_len {} clamp_len {}'.format(
|
| 97 |
+
args.batch_size, args.tgt_len, args.ext_len, args.mem_len, args.clamp_len))
|
| 98 |
+
|
| 99 |
+
model.reset_length(args.tgt_len, args.ext_len, args.mem_len)
|
| 100 |
+
if args.clamp_len > 0:
|
| 101 |
+
model.clamp_len = args.clamp_len
|
| 102 |
+
if args.same_length:
|
| 103 |
+
model.same_length = True
|
| 104 |
+
|
| 105 |
+
###############################################################################
|
| 106 |
+
# Evaluation code
|
| 107 |
+
###############################################################################
|
| 108 |
+
def evaluate(eval_iter):
|
| 109 |
+
# Turn on evaluation mode which disables dropout.
|
| 110 |
+
model.eval()
|
| 111 |
+
total_len, total_loss = 0, 0.
|
| 112 |
+
start_time = time.time()
|
| 113 |
+
with torch.no_grad():
|
| 114 |
+
mems = None
|
| 115 |
+
for idx, (data, target, seq_len) in enumerate(eval_iter):
|
| 116 |
+
ret = model(data, target, mems)
|
| 117 |
+
loss, mems = ret
|
| 118 |
+
loss = loss.mean()
|
| 119 |
+
total_loss += seq_len * loss.item()
|
| 120 |
+
total_len += seq_len
|
| 121 |
+
total_time = time.time() - start_time
|
| 122 |
+
logger.info('Time : {:.2f}s, {:.2f}ms/segment'.format(
|
| 123 |
+
total_time, 1000 * total_time / (idx+1)))
|
| 124 |
+
return total_loss / total_len
|
| 125 |
+
|
| 126 |
+
# Run on test data.
|
| 127 |
+
if args.split == 'all':
|
| 128 |
+
test_loss = evaluate(te_iter)
|
| 129 |
+
valid_loss = evaluate(va_iter)
|
| 130 |
+
elif args.split == 'valid':
|
| 131 |
+
valid_loss = evaluate(va_iter)
|
| 132 |
+
test_loss = None
|
| 133 |
+
elif args.split == 'test':
|
| 134 |
+
test_loss = evaluate(te_iter)
|
| 135 |
+
valid_loss = None
|
| 136 |
+
|
| 137 |
+
def format_log(loss, split):
|
| 138 |
+
log_str = '| {0} loss {1:5.2f} | {0} ppl {2:9.3f} '.format(
|
| 139 |
+
split, loss, math.exp(loss))
|
| 140 |
+
return log_str
|
| 141 |
+
|
| 142 |
+
log_str = ''
|
| 143 |
+
if valid_loss is not None:
|
| 144 |
+
log_str += format_log(valid_loss, 'valid')
|
| 145 |
+
if test_loss is not None:
|
| 146 |
+
log_str += format_log(test_loss, 'test')
|
| 147 |
+
|
| 148 |
+
logger.info('=' * 100)
|
| 149 |
+
logger.info(log_str)
|
| 150 |
+
logger.info('=' * 100)
|
| 151 |
+
|
| 152 |
+
if __name__ == '__main__':
|
| 153 |
+
main()
|
dark-secrets-of-BERT-master/dark-secrets-of-BERT-master/hubconf.py
ADDED
|
@@ -0,0 +1,187 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
from pytorch_pretrained_bert.tokenization import BertTokenizer
|
| 2 |
+
from pytorch_pretrained_bert.modeling import (
|
| 3 |
+
BertModel,
|
| 4 |
+
BertForNextSentencePrediction,
|
| 5 |
+
BertForMaskedLM,
|
| 6 |
+
BertForMultipleChoice,
|
| 7 |
+
BertForPreTraining,
|
| 8 |
+
BertForQuestionAnswering,
|
| 9 |
+
BertForSequenceClassification,
|
| 10 |
+
BertForTokenClassification,
|
| 11 |
+
)
|
| 12 |
+
|
| 13 |
+
dependencies = ['torch', 'tqdm', 'boto3', 'requests', 'regex']
|
| 14 |
+
|
| 15 |
+
# A lot of models share the same param doc. Use a decorator
|
| 16 |
+
# to save typing
|
| 17 |
+
bert_docstring = """
|
| 18 |
+
Params:
|
| 19 |
+
pretrained_model_name_or_path: either:
|
| 20 |
+
- a str with the name of a pre-trained model to load
|
| 21 |
+
. `bert-base-uncased`
|
| 22 |
+
. `bert-large-uncased`
|
| 23 |
+
. `bert-base-cased`
|
| 24 |
+
. `bert-large-cased`
|
| 25 |
+
. `bert-base-multilingual-uncased`
|
| 26 |
+
. `bert-base-multilingual-cased`
|
| 27 |
+
. `bert-base-chinese`
|
| 28 |
+
- a path or url to a pretrained model archive containing:
|
| 29 |
+
. `bert_config.json` a configuration file for the model
|
| 30 |
+
. `pytorch_model.bin` a PyTorch dump of a BertForPreTraining
|
| 31 |
+
instance
|
| 32 |
+
- a path or url to a pretrained model archive containing:
|
| 33 |
+
. `bert_config.json` a configuration file for the model
|
| 34 |
+
. `model.chkpt` a TensorFlow checkpoint
|
| 35 |
+
from_tf: should we load the weights from a locally saved TensorFlow
|
| 36 |
+
checkpoint
|
| 37 |
+
cache_dir: an optional path to a folder in which the pre-trained models
|
| 38 |
+
will be cached.
|
| 39 |
+
state_dict: an optional state dictionnary
|
| 40 |
+
(collections.OrderedDict object) to use instead of Google
|
| 41 |
+
pre-trained models
|
| 42 |
+
*inputs, **kwargs: additional input for the specific Bert class
|
| 43 |
+
(ex: num_labels for BertForSequenceClassification)
|
| 44 |
+
"""
|
| 45 |
+
|
| 46 |
+
|
| 47 |
+
def _append_from_pretrained_docstring(docstr):
|
| 48 |
+
def docstring_decorator(fn):
|
| 49 |
+
fn.__doc__ = fn.__doc__ + docstr
|
| 50 |
+
return fn
|
| 51 |
+
return docstring_decorator
|
| 52 |
+
|
| 53 |
+
|
| 54 |
+
def bertTokenizer(*args, **kwargs):
|
| 55 |
+
"""
|
| 56 |
+
Instantiate a BertTokenizer from a pre-trained/customized vocab file
|
| 57 |
+
Args:
|
| 58 |
+
pretrained_model_name_or_path: Path to pretrained model archive
|
| 59 |
+
or one of pre-trained vocab configs below.
|
| 60 |
+
* bert-base-uncased
|
| 61 |
+
* bert-large-uncased
|
| 62 |
+
* bert-base-cased
|
| 63 |
+
* bert-large-cased
|
| 64 |
+
* bert-base-multilingual-uncased
|
| 65 |
+
* bert-base-multilingual-cased
|
| 66 |
+
* bert-base-chinese
|
| 67 |
+
Keyword args:
|
| 68 |
+
cache_dir: an optional path to a specific directory to download and cache
|
| 69 |
+
the pre-trained model weights.
|
| 70 |
+
Default: None
|
| 71 |
+
do_lower_case: Whether to lower case the input.
|
| 72 |
+
Only has an effect when do_wordpiece_only=False
|
| 73 |
+
Default: True
|
| 74 |
+
do_basic_tokenize: Whether to do basic tokenization before wordpiece.
|
| 75 |
+
Default: True
|
| 76 |
+
max_len: An artificial maximum length to truncate tokenized sequences to;
|
| 77 |
+
Effective maximum length is always the minimum of this
|
| 78 |
+
value (if specified) and the underlying BERT model's
|
| 79 |
+
sequence length.
|
| 80 |
+
Default: None
|
| 81 |
+
never_split: List of tokens which will never be split during tokenization.
|
| 82 |
+
Only has an effect when do_wordpiece_only=False
|
| 83 |
+
Default: ["[UNK]", "[SEP]", "[PAD]", "[CLS]", "[MASK]"]
|
| 84 |
+
|
| 85 |
+
Example:
|
| 86 |
+
>>> sentence = 'Hello, World!'
|
| 87 |
+
>>> tokenizer = torch.hub.load('ailzhang/pytorch-pretrained-BERT:hubconf', 'bertTokenizer', 'bert-base-cased', do_basic_tokenize=False, force_reload=False)
|
| 88 |
+
>>> toks = tokenizer.tokenize(sentence)
|
| 89 |
+
['Hello', '##,', 'World', '##!']
|
| 90 |
+
>>> ids = tokenizer.convert_tokens_to_ids(toks)
|
| 91 |
+
[8667, 28136, 1291, 28125]
|
| 92 |
+
"""
|
| 93 |
+
tokenizer = BertTokenizer.from_pretrained(*args, **kwargs)
|
| 94 |
+
return tokenizer
|
| 95 |
+
|
| 96 |
+
|
| 97 |
+
@_append_from_pretrained_docstring(bert_docstring)
|
| 98 |
+
def bertModel(*args, **kwargs):
|
| 99 |
+
"""
|
| 100 |
+
BertModel is the basic BERT Transformer model with a layer of summed token,
|
| 101 |
+
position and sequence embeddings followed by a series of identical
|
| 102 |
+
self-attention blocks (12 for BERT-base, 24 for BERT-large).
|
| 103 |
+
"""
|
| 104 |
+
model = BertModel.from_pretrained(*args, **kwargs)
|
| 105 |
+
return model
|
| 106 |
+
|
| 107 |
+
|
| 108 |
+
@_append_from_pretrained_docstring(bert_docstring)
|
| 109 |
+
def bertForNextSentencePrediction(*args, **kwargs):
|
| 110 |
+
"""
|
| 111 |
+
BERT model with next sentence prediction head.
|
| 112 |
+
This module comprises the BERT model followed by the next sentence
|
| 113 |
+
classification head.
|
| 114 |
+
"""
|
| 115 |
+
model = BertForNextSentencePrediction.from_pretrained(*args, **kwargs)
|
| 116 |
+
return model
|
| 117 |
+
|
| 118 |
+
|
| 119 |
+
@_append_from_pretrained_docstring(bert_docstring)
|
| 120 |
+
def bertForPreTraining(*args, **kwargs):
|
| 121 |
+
"""
|
| 122 |
+
BERT model with pre-training heads.
|
| 123 |
+
This module comprises the BERT model followed by the two pre-training heads
|
| 124 |
+
- the masked language modeling head, and
|
| 125 |
+
- the next sentence classification head.
|
| 126 |
+
"""
|
| 127 |
+
model = BertForPreTraining.from_pretrained(*args, **kwargs)
|
| 128 |
+
return model
|
| 129 |
+
|
| 130 |
+
|
| 131 |
+
@_append_from_pretrained_docstring(bert_docstring)
|
| 132 |
+
def bertForMaskedLM(*args, **kwargs):
|
| 133 |
+
"""
|
| 134 |
+
BertForMaskedLM includes the BertModel Transformer followed by the
|
| 135 |
+
(possibly) pre-trained masked language modeling head.
|
| 136 |
+
"""
|
| 137 |
+
model = BertForMaskedLM.from_pretrained(*args, **kwargs)
|
| 138 |
+
return model
|
| 139 |
+
|
| 140 |
+
|
| 141 |
+
@_append_from_pretrained_docstring(bert_docstring)
|
| 142 |
+
def bertForSequenceClassification(*args, **kwargs):
|
| 143 |
+
"""
|
| 144 |
+
BertForSequenceClassification is a fine-tuning model that includes
|
| 145 |
+
BertModel and a sequence-level (sequence or pair of sequences) classifier
|
| 146 |
+
on top of the BertModel.
|
| 147 |
+
|
| 148 |
+
The sequence-level classifier is a linear layer that takes as input the
|
| 149 |
+
last hidden state of the first character in the input sequence
|
| 150 |
+
(see Figures 3a and 3b in the BERT paper).
|
| 151 |
+
"""
|
| 152 |
+
model = BertForSequenceClassification.from_pretrained(*args, **kwargs)
|
| 153 |
+
return model
|
| 154 |
+
|
| 155 |
+
|
| 156 |
+
@_append_from_pretrained_docstring(bert_docstring)
|
| 157 |
+
def bertForMultipleChoice(*args, **kwargs):
|
| 158 |
+
"""
|
| 159 |
+
BertForMultipleChoice is a fine-tuning model that includes BertModel and a
|
| 160 |
+
linear layer on top of the BertModel.
|
| 161 |
+
"""
|
| 162 |
+
model = BertForMultipleChoice.from_pretrained(*args, **kwargs)
|
| 163 |
+
return model
|
| 164 |
+
|
| 165 |
+
|
| 166 |
+
@_append_from_pretrained_docstring(bert_docstring)
|
| 167 |
+
def bertForQuestionAnswering(*args, **kwargs):
|
| 168 |
+
"""
|
| 169 |
+
BertForQuestionAnswering is a fine-tuning model that includes BertModel
|
| 170 |
+
with a token-level classifiers on top of the full sequence of last hidden
|
| 171 |
+
states.
|
| 172 |
+
"""
|
| 173 |
+
model = BertForQuestionAnswering.from_pretrained(*args, **kwargs)
|
| 174 |
+
return model
|
| 175 |
+
|
| 176 |
+
|
| 177 |
+
@_append_from_pretrained_docstring(bert_docstring)
|
| 178 |
+
def bertForTokenClassification(*args, **kwargs):
|
| 179 |
+
"""
|
| 180 |
+
BertForTokenClassification is a fine-tuning model that includes BertModel
|
| 181 |
+
and a token-level classifier on top of the BertModel.
|
| 182 |
+
|
| 183 |
+
The token-level classifier is a linear layer that takes as input the last
|
| 184 |
+
hidden state of the sequence.
|
| 185 |
+
"""
|
| 186 |
+
model = BertForTokenClassification.from_pretrained(*args, **kwargs)
|
| 187 |
+
return model
|
dark-secrets-of-BERT-master/dark-secrets-of-BERT-master/notebooks/Comparing-TF-and-PT-models-MLM-NSP.ipynb
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|